1 00:00:06,220 --> 00:00:09,580 It is probably the most amazing thing you do 2 00:00:09,580 --> 00:00:11,900 and it defines you as being human. 3 00:00:11,900 --> 00:00:16,620 And you do it with ease, so much so you barely notice you're doing it. 4 00:00:18,180 --> 00:00:22,540 Apple! Yeah, apple! 5 00:00:22,540 --> 00:00:24,940 THEY SPEAK IN NATIVE TONGUE 6 00:00:43,780 --> 00:00:46,220 Your ability to talk. 7 00:00:46,220 --> 00:00:49,180 Or more precisely, the way you use language. 8 00:00:51,220 --> 00:00:54,220 THEY SPEAK IN NATIVE TONGUE 9 00:00:56,780 --> 00:00:57,820 Language. 10 00:01:10,180 --> 00:01:12,860 COMPUTERISED VOICE: Hi, baby! 11 00:01:12,860 --> 00:01:16,820 From the moment we're born, we have language in our lives. 12 00:01:16,820 --> 00:01:19,100 Hi, baby! 13 00:01:19,100 --> 00:01:23,300 A unique ability that defines us as human. 14 00:01:23,300 --> 00:01:27,380 I can walk up to someone I don't know 15 00:01:27,380 --> 00:01:31,460 and I can make a sequence of noises that I've never made before 16 00:01:31,460 --> 00:01:33,660 by pushing air through my mouth. 17 00:01:35,380 --> 00:01:39,620 I will take a thought in my head and make it go into their head, 18 00:01:39,620 --> 00:01:41,900 and that's an incredible trick. 19 00:01:44,540 --> 00:01:47,780 Other animals may use sounds to communicate 20 00:01:47,780 --> 00:01:51,420 but we're the only ones on the planet who can talk. 21 00:01:51,420 --> 00:01:58,540 Speech and language distinguishes us from all other species. 22 00:01:58,540 --> 00:02:04,380 We are not only using words to be able to express ourselves 23 00:02:04,380 --> 00:02:07,500 but we are really expressing our thought processes 24 00:02:07,500 --> 00:02:09,300 that are unique to ourselves. 25 00:02:10,780 --> 00:02:15,820 For most of us it just happens. 26 00:02:15,820 --> 00:02:18,580 It's such a sophisticated skill 27 00:02:18,580 --> 00:02:23,980 and yet children learn it so easily with minimal effort at all. 28 00:02:23,980 --> 00:02:26,300 That's Mummy. 29 00:02:26,300 --> 00:02:28,740 And when you start thinking about it, 30 00:02:28,740 --> 00:02:31,620 it is quite miraculous how the brain does it. 31 00:02:31,620 --> 00:02:37,180 But despite decades of research, how we learn to talk remains a mystery. 32 00:02:37,180 --> 00:02:39,860 How did this ability evolve? 33 00:02:39,860 --> 00:02:41,660 Why is it uniquely human? 34 00:02:41,660 --> 00:02:46,860 And is it something we are born with or something that we learn? 35 00:02:46,860 --> 00:02:49,300 It strikes me it's surprising and odd 36 00:02:49,300 --> 00:02:52,460 that we know an enormous amount about, for example, 37 00:02:52,460 --> 00:02:56,460 the first fractions of a second in the history of the universe, 38 00:02:56,460 --> 00:03:00,060 but we really know very little at all about what makes us human 39 00:03:00,060 --> 00:03:01,940 and where language comes from. 40 00:03:09,580 --> 00:03:11,900 BABY BABBLES 41 00:03:19,460 --> 00:03:20,780 Daddy. 44 00:03:34,420 --> 00:03:37,060 But how we start to talk is now being observed 45 00:03:37,060 --> 00:03:39,260 minute by minute for the first time. 46 00:03:43,940 --> 00:03:46,980 Cognitive scientist Deb Roy realised 47 00:03:46,980 --> 00:03:50,700 that to discover how a child learns to speak... 48 00:03:50,700 --> 00:03:54,940 He'd have to be able to see and hear them 24 hours a day. 49 00:03:56,500 --> 00:03:59,980 His solution was to take the extraordinary step 50 00:03:59,980 --> 00:04:03,500 of turning his own home into a language laboratory, 51 00:04:03,500 --> 00:04:05,980 and his guinea pig was his own son. 52 00:04:08,100 --> 00:04:10,820 My wife and I were expecting our first child 53 00:04:10,820 --> 00:04:16,180 and we had many conversations around, you know, would this be possible, 54 00:04:16,180 --> 00:04:19,500 could we set up our home as an observation space 55 00:04:19,500 --> 00:04:23,780 and actually capture this kind of rich longitudinal record 56 00:04:23,780 --> 00:04:27,020 of our own child's development from birth? 57 00:04:27,020 --> 00:04:29,060 And that's what we set out to do. 58 00:04:34,700 --> 00:04:37,700 And so was born the Speech Home Project, 59 00:04:39,900 --> 00:04:44,260 The most ambitious language observation experiment ever seen. 60 00:04:44,260 --> 00:04:48,500 Cameras and microphones recorded every second of Deb's son's life 61 00:04:48,500 --> 00:04:52,300 from the moment he was born until he was three-years-old. 62 00:04:54,940 --> 00:05:00,340 Capturing in precise detail the process of how we learn to talk. 63 00:05:02,020 --> 00:05:03,060 Baa! 64 00:05:03,060 --> 00:05:05,220 Ball. 65 00:05:07,140 --> 00:05:09,740 Blue ball. Blue ball. 66 00:05:09,740 --> 00:05:12,140 Blue ball, yeah. 67 00:05:12,140 --> 00:05:15,460 From the earliest unintelligible babble 68 00:05:15,460 --> 00:05:18,340 to the emergence of a very first word. 69 00:05:21,940 --> 00:05:24,020 Apple. Apple. 70 00:05:24,020 --> 00:05:27,140 Yeah, apple. 71 00:05:27,420 --> 00:05:30,420 My interest going in was really to understand 72 00:05:30,420 --> 00:05:33,380 the earliest stages of language formation, 73 00:05:33,380 --> 00:05:36,700 and the typical development stages in language, 74 00:05:36,700 --> 00:05:40,100 a child will go from babbling to using single words 75 00:05:40,100 --> 00:05:43,420 usually as a complete request and descriptions, 76 00:05:43,420 --> 00:05:47,820 and then from that they will start what's called the two-word stage 77 00:05:47,820 --> 00:05:51,500 where they'll put two words together like "more milk" 78 00:05:51,500 --> 00:05:56,420 and from that, very quickly more complex grammatical structures emerge. 79 00:06:04,220 --> 00:06:07,220 My focus and my aim was to capture the phase 80 00:06:07,220 --> 00:06:09,220 up to the two-word utterance 81 00:06:09,220 --> 00:06:13,980 and that could happen anywhere between second and third birthday. 82 00:06:13,980 --> 00:06:16,100 It turned out my son was an early talker 83 00:06:16,100 --> 00:06:18,300 so by the time his second birthday arrived 84 00:06:18,300 --> 00:06:20,100 we had the main data set we wanted. 85 00:06:21,620 --> 00:06:24,580 By the time recording was complete, 86 00:06:24,580 --> 00:06:27,820 more than 240,000 hours of information 87 00:06:27,820 --> 00:06:31,180 and 16 million words had been collected. 88 00:06:31,180 --> 00:06:36,060 It's a lot of data but in its raw form it's useless 89 00:06:36,060 --> 00:06:38,860 and so the challenges this now sets up for us 90 00:06:38,860 --> 00:06:42,700 is how do you start extracting the right kind of metadata, 91 00:06:42,700 --> 00:06:44,740 transcripts of who said what, 92 00:06:44,740 --> 00:06:47,460 annotations of where those people were, 93 00:06:47,460 --> 00:06:49,860 annotations of how they're moving 94 00:06:49,860 --> 00:06:54,020 and the relationships that they were in as they were speaking. 95 00:06:54,020 --> 00:06:57,100 And these are the, the tools that we are now building 96 00:06:57,100 --> 00:06:58,580 to analyse the raw data, 97 00:06:58,580 --> 00:07:01,060 and from that, we're starting to see some, 98 00:07:01,060 --> 00:07:04,900 some early insights into the patterns of language development. 99 00:07:08,660 --> 00:07:11,060 Despite the mountains of video footage, 100 00:07:11,060 --> 00:07:13,980 Deb Roy is already uncovering the intricate effect 101 00:07:13,980 --> 00:07:16,300 daily life has on language development. 102 00:07:17,860 --> 00:07:22,380 It seems cooing to your baby is not just instinct, it's essential. 103 00:07:22,380 --> 00:07:25,740 In the months leading up to their son's first words, 104 00:07:25,740 --> 00:07:29,420 he and his wife unconsciously simplified their speech. 105 00:07:31,500 --> 00:07:34,100 Then as their son's own speech develops, 106 00:07:34,100 --> 00:07:36,740 they began to use longer sentences again, 107 00:07:36,740 --> 00:07:39,300 mirroring the child's own development. 108 00:07:39,300 --> 00:07:43,220 What's that over there? Over there, that ball? 109 00:07:43,220 --> 00:07:47,900 Green ball. Oh! 110 00:07:47,900 --> 00:07:51,860 For the very first time, the analysis has also enabled 111 00:07:51,860 --> 00:07:57,020 Deb to piece together the precise emergence of individual words, 112 00:07:57,020 --> 00:07:59,580 charting how a new word is born. 113 00:07:59,580 --> 00:08:01,740 OK, water, water. 114 00:08:01,740 --> 00:08:03,740 Gaga, gaga. 115 00:08:03,740 --> 00:08:06,500 So my son, around his first birthday, 116 00:08:06,500 --> 00:08:09,940 started using the sound "gaga" to indicate water, 117 00:08:09,940 --> 00:08:12,420 and then over the next several months 118 00:08:12,420 --> 00:08:16,020 learned to slowly approximate the proper speech form. 119 00:08:16,020 --> 00:08:17,980 Water. 120 00:08:17,980 --> 00:08:22,140 Just like time lapse video lets you capture a flower blossoming, 121 00:08:22,140 --> 00:08:27,020 what we're going to hear is the blossoming of a speech form. 122 00:08:27,020 --> 00:08:28,140 OK, water? 123 00:08:28,140 --> 00:08:29,700 Water. 124 00:08:36,420 --> 00:08:39,100 Water? Water? 125 00:08:40,220 --> 00:08:41,300 Water. 126 00:08:46,940 --> 00:08:48,460 That's water. 127 00:09:00,420 --> 00:09:05,580 It will take years for Deb Roy and his team to analyse all of the data 128 00:09:05,580 --> 00:09:07,740 but buried within this footage 129 00:09:07,740 --> 00:09:10,580 are the secrets of how we learn language. 130 00:09:10,580 --> 00:09:12,620 If the patterns can be unlocked, 131 00:09:12,620 --> 00:09:15,780 then once this remarkable project is complete 132 00:09:15,780 --> 00:09:18,620 we will have the clearest understanding yet 133 00:09:18,620 --> 00:09:22,740 of how a child achieves the remarkable feat of learning to talk. 134 00:09:33,900 --> 00:09:38,820 By the time a child is five, they'll know as many as 5,000 words. 135 00:09:38,820 --> 00:09:40,900 THEY SPEAKS IN NATIVE TONGUE 136 00:09:58,060 --> 00:10:00,940 My voice changes when I get in a situation. 137 00:10:00,940 --> 00:10:04,820 That requires a rather precise placement of the tip of the tongue. 138 00:10:04,820 --> 00:10:07,580 At the third stroke the time will be five... 139 00:10:07,580 --> 00:10:09,140 Language is exclusively human 140 00:10:09,140 --> 00:10:11,700 and it's something that comes naturally to us. 141 00:10:11,700 --> 00:10:17,420 From their voice, so it's clear the voice is deeply expressive. 142 00:10:17,420 --> 00:10:20,660 Despite decades of intense investigation, 143 00:10:20,660 --> 00:10:23,940 we still know very little about the basic origins of language. 144 00:10:23,940 --> 00:10:25,700 ELEPHANT ROARS 145 00:10:25,700 --> 00:10:26,900 Sunlight. 146 00:10:26,900 --> 00:10:29,420 While other creatures can communicate 147 00:10:29,420 --> 00:10:32,300 their basic needs and emotions to each other, 148 00:10:32,300 --> 00:10:38,900 we're the only species that can convey complex meaning and thought. 149 00:10:38,900 --> 00:10:43,380 T or the D at the end of a word for, say, dot or dog. 150 00:10:47,260 --> 00:10:51,540 It's remarkable that no other animal has ever developed speech. 151 00:10:51,540 --> 00:10:56,180 What is it that makes us unique? 152 00:10:59,220 --> 00:11:00,700 Do this, Vicky. 153 00:11:00,700 --> 00:11:04,300 HE BLOWS RASPBERRY 154 00:11:04,300 --> 00:11:07,180 Another sound resembles the letter K. 155 00:11:07,180 --> 00:11:08,780 Vicky, sit up. 156 00:11:08,780 --> 00:11:12,140 Not even our closest relatives, the chimps, can talk. 157 00:11:12,140 --> 00:11:14,700 Psychologists spent nearly seven years 158 00:11:14,700 --> 00:11:17,780 trying to teach Vicky to speak with little success. 159 00:11:17,780 --> 00:11:20,180 Resembles the letter T. 160 00:11:22,060 --> 00:11:26,620 And in this case they really tried to teach a chimp to say some basic words 161 00:11:26,620 --> 00:11:29,620 and completely failed, and everyone knew of course that, 162 00:11:29,620 --> 00:11:32,060 say, parrots can learn to imitate speech. 163 00:11:32,060 --> 00:11:35,500 So it was kind of a very surprising and eye-opening experience 164 00:11:35,500 --> 00:11:40,060 to see that these chimps with great big brains for an animal were not able to do this. 165 00:11:40,060 --> 00:11:42,420 It was already very clear that chimpanzees 166 00:11:42,420 --> 00:11:45,820 don't have the kind of control over their vocal tract that we do. 167 00:11:47,660 --> 00:11:51,780 So once people knew that chimpanzees couldn't produce speech, 168 00:11:51,780 --> 00:11:53,460 the big question was why? 169 00:12:01,860 --> 00:12:04,740 Scientists believed they knew the answer. 170 00:12:04,740 --> 00:12:07,980 It was all down to our unique anatomy. 171 00:12:07,980 --> 00:12:13,660 We have a larynx, or voice box, that's fixed low down in our throat. 172 00:12:13,660 --> 00:12:16,180 This makes the vocal tract longer 173 00:12:16,180 --> 00:12:21,420 and gives us the physical ability to produce a wide variety of sounds. 174 00:12:21,420 --> 00:12:25,780 The voice box of all other animals was found to be high in the throat, 175 00:12:25,780 --> 00:12:30,020 leaving them incapable of making the complicated sounds of speech 176 00:12:30,020 --> 00:12:31,940 in their short vocal tract. 177 00:12:31,940 --> 00:12:38,180 But nobody had actually ever seen inside the throat of an animal 178 00:12:38,180 --> 00:12:39,900 as it vocalised. 179 00:12:43,980 --> 00:12:46,780 Tecumseh Fitch had a unique piece of equipment 180 00:12:46,780 --> 00:12:49,500 to solve the problem, a video X-ray machine 181 00:12:49,500 --> 00:12:52,860 that had been developed to study animals swallowing. 182 00:12:52,860 --> 00:12:54,900 What we found was really astounding. 183 00:12:54,900 --> 00:12:58,460 It wasn't anything we were prepared for by reading the literature 184 00:12:58,460 --> 00:13:00,300 and here's what we saw. 185 00:13:06,300 --> 00:13:10,820 So what this is is the larynx pulling down out of the nasal cavity 186 00:13:10,820 --> 00:13:13,260 so that during the vocalisation 187 00:13:13,260 --> 00:13:16,700 the animal was vocalising out through its mouth. 188 00:13:18,300 --> 00:13:22,140 The image made no sense so Fitch looked at other mammals, 189 00:13:22,140 --> 00:13:23,860 including Megan the dog. 190 00:13:28,180 --> 00:13:31,620 So you can see the dog's jaw and the tongue. 191 00:13:31,620 --> 00:13:33,300 This is the larynx right here. 192 00:13:36,180 --> 00:13:40,300 And you can see that with each bark the larynx kind of jerks down, 193 00:13:40,300 --> 00:13:41,940 so let's slow that down. 194 00:13:44,340 --> 00:13:46,060 So here's the larynx. 195 00:13:47,620 --> 00:13:51,900 You can see that it's really a quite pronounced descent 196 00:13:51,900 --> 00:13:55,420 of the larynx during the barks. It pulls down. 197 00:13:56,980 --> 00:14:00,620 What Fitch discovered has revolutionised our understanding 198 00:14:00,620 --> 00:14:01,780 of why we can talk. 199 00:14:04,940 --> 00:14:08,140 The mammalian vocal tract is incredibly flexible 200 00:14:08,140 --> 00:14:11,940 so whether we're looking at dogs or pigs or goats or monkeys, 201 00:14:11,940 --> 00:14:15,060 we always see the same thing, that the, the anatomy 202 00:14:15,060 --> 00:14:19,500 rearranges itself very dynamically while the animal is vocalising. 203 00:14:20,700 --> 00:14:23,700 So it seems that if it was just down to anatomy, 204 00:14:23,700 --> 00:14:26,580 all mammals would have the ability to talk. 205 00:14:29,060 --> 00:14:33,220 The main conclusion that I draw from all this work is that the constraint 206 00:14:33,220 --> 00:14:35,780 that's keeping a chimpanzee from speaking, 207 00:14:35,780 --> 00:14:38,140 or indeed keeping a dog from speaking, 208 00:14:38,140 --> 00:14:40,060 is not the peripheral vocal anatomy, 209 00:14:40,060 --> 00:14:44,060 because any of these animals that we've looked at are able to lower the larynx 210 00:14:44,060 --> 00:14:47,340 and reconfigure the vocal tract into a human-like confirmation 211 00:14:47,340 --> 00:14:49,180 basically any time they vocalise. 212 00:14:49,180 --> 00:14:52,820 So that can't be the main thing that's keeping these other animals 213 00:14:52,820 --> 00:14:55,820 from having the capacity to produce humanlike speech, 214 00:14:55,820 --> 00:14:58,780 so it must be in the brain, by process of elimination. 215 00:14:58,780 --> 00:15:01,060 Come on, ?4 a pound, banana. 216 00:15:01,060 --> 00:15:04,420 Yes, I had to speak to both those... 217 00:15:04,420 --> 00:15:07,980 Every dish has sugar. 218 00:15:07,980 --> 00:15:14,100 We don't stop to think how our brain processes the 370 million words we will say in our lifetime. 219 00:15:14,100 --> 00:15:17,060 Read my lips. 220 00:15:17,060 --> 00:15:20,780 Or how we coordinate the complex sequence of thoughts, 221 00:15:20,780 --> 00:15:25,340 movements and actions that lie behind each and every word we utter. 222 00:15:25,340 --> 00:15:28,980 But if we are to ever truly understand language, 223 00:15:28,980 --> 00:15:31,420 then it is deep inside our heads 224 00:15:31,420 --> 00:15:34,660 where some of the greatest mysteries lie. 225 00:15:41,140 --> 00:15:42,660 The more we uncover, 226 00:15:42,660 --> 00:15:47,900 the more we realise how delicate and fragile these processes can be. 227 00:15:47,900 --> 00:15:50,700 In 2006 Steve Steere was in an accident 228 00:15:50,700 --> 00:15:53,420 that would change his life for ever. 229 00:15:56,420 --> 00:15:59,300 Steve was always a very active person. 230 00:15:59,300 --> 00:16:02,380 He had done two sponsored rides, 231 00:16:02,380 --> 00:16:06,580 one to Mexico and one to Ecuador, which he thoroughly enjoyed 232 00:16:06,580 --> 00:16:11,220 and he was training for the third ride, which was to go to Peru. 233 00:16:12,780 --> 00:16:15,900 When he had an accident, 234 00:16:15,900 --> 00:16:21,980 he was found unconscious and he was taken to hospital, 235 00:16:21,980 --> 00:16:26,380 where they diagnosed him with a fractured shoulder. 236 00:16:29,500 --> 00:16:33,180 Initially, Steve's injuries seemed relatively minor. 237 00:16:35,460 --> 00:16:40,260 But early the next morning things took a turn for the worse 238 00:16:40,260 --> 00:16:42,580 and he had a massive stroke. 239 00:16:42,580 --> 00:16:45,300 This left him with severe brain damage. 240 00:16:45,300 --> 00:16:47,940 He was told he'd never talk properly again. 241 00:16:49,620 --> 00:16:51,540 I remember watching a programme 242 00:16:51,540 --> 00:16:54,300 probably six months before Steve had his accident, 243 00:16:54,300 --> 00:16:57,220 which showed a lady caring for her disabled husband 244 00:16:57,220 --> 00:16:58,580 and I remember feeling, 245 00:16:58,580 --> 00:17:02,420 "Gosh, if that happened to me, I don't think I'd be able to do that." 246 00:17:02,420 --> 00:17:08,060 I remember sitting for about two hours in the room when he was asleep, 247 00:17:08,060 --> 00:17:11,540 raging inside that this was not fair. 248 00:17:11,540 --> 00:17:16,940 We'd just got our business going and it was doing really well, 249 00:17:16,940 --> 00:17:19,580 and we had two young children 250 00:17:19,580 --> 00:17:22,780 and I just think sometimes it's not fair. 251 00:17:37,820 --> 00:17:40,820 Cases like Steve's offer scientists like Cathy Price 252 00:17:40,820 --> 00:17:44,460 a unique window into how the brain and language connect together. 253 00:17:46,260 --> 00:17:50,340 20 years ago I was told that the brain wasn't relevant 254 00:17:50,340 --> 00:17:52,220 to our study of language 255 00:17:52,220 --> 00:17:56,980 and the reason for that was that we had so little information. 256 00:17:56,980 --> 00:18:01,420 The two types of information that we had were either from autopsies 257 00:18:01,420 --> 00:18:05,660 of people who had died after they had had a language difficulty, 258 00:18:05,660 --> 00:18:09,940 or we were using brain scans where the information was so blurred 259 00:18:09,940 --> 00:18:14,780 that you couldn't really see very much detail about what was happening. 260 00:18:14,780 --> 00:18:16,780 So at that point all we knew was that 261 00:18:16,780 --> 00:18:20,020 the left side of the brain was most important for language 262 00:18:20,020 --> 00:18:23,300 and the front part of the brain is important for speaking 263 00:18:23,300 --> 00:18:27,260 and the back part of the brain was important for understanding speech 264 00:18:27,260 --> 00:18:30,940 so that was the sort of general level at which it was understood. 265 00:18:30,940 --> 00:18:33,340 As scanning technology has improved, 266 00:18:33,340 --> 00:18:37,260 it's become apparent just how complicated the brain functions 267 00:18:37,260 --> 00:18:39,460 involved in language really are. 268 00:18:39,460 --> 00:18:42,740 Hello, Steve, can you hear me? That's great. 269 00:18:42,740 --> 00:18:46,020 To build a complete picture of which parts of the brain 270 00:18:46,020 --> 00:18:48,300 are involved in language in all of us, 271 00:18:48,300 --> 00:18:51,700 Cathy Price matches the exact nature of a speech defect 272 00:18:51,700 --> 00:18:54,540 with the exact position of damage in the brain. 273 00:18:54,540 --> 00:18:56,780 When I look at Steve's brain here, 274 00:18:56,780 --> 00:19:00,500 I can see that he's lost a very extensive set of regions 275 00:19:00,500 --> 00:19:04,140 in his left hemisphere that are important for speech 276 00:19:04,140 --> 00:19:06,540 and he can still produce some speech 277 00:19:06,540 --> 00:19:09,460 but it will be extremely difficult for him. 278 00:19:11,020 --> 00:19:14,300 A fully-fledged ability to communicate with language 279 00:19:14,300 --> 00:19:18,100 relies on many separate processes working together within us. 280 00:19:18,100 --> 00:19:20,580 We need to use our memory, our senses 281 00:19:20,580 --> 00:19:24,620 and have precise motor control over our mouths and tongues. 282 00:19:26,740 --> 00:19:28,420 What I'd like you to do 283 00:19:28,420 --> 00:19:33,020 is to point to the picture that matches with the words, OK? 284 00:19:33,020 --> 00:19:37,220 So the words are, "the pen is under the paper." 285 00:19:37,220 --> 00:19:41,140 Cathy needs to identify the problems Steve has with speaking 286 00:19:41,140 --> 00:19:45,020 before she can link them back to the precise areas in his brain. 287 00:19:45,020 --> 00:19:46,980 This one. That's right. 288 00:19:46,980 --> 00:19:51,180 When I ask him to point to pictures of particular scenes, 289 00:19:51,180 --> 00:19:54,500 he could pick out which scene matched the sentence 290 00:19:54,500 --> 00:19:58,300 and that's quite a difficult thing to do if you can't speak 291 00:19:58,300 --> 00:20:01,860 because you can't hear the words go round in your head. 292 00:20:01,860 --> 00:20:04,780 You can't repeat it to yourself. 293 00:20:04,780 --> 00:20:07,060 So he did very well on that task. 294 00:20:07,060 --> 00:20:13,300 The butcher shoots the nurse. Um. 295 00:20:14,500 --> 00:20:16,340 It's a bit violent. 296 00:20:16,340 --> 00:20:19,260 Yes. This one's a nicer one. 297 00:20:19,260 --> 00:20:22,780 One crucial area of Steve's brain is still functioning. 298 00:20:22,780 --> 00:20:25,860 The anterior superior temporal sulcus is involved 299 00:20:25,860 --> 00:20:28,300 in understanding the meaning of words, 300 00:20:28,300 --> 00:20:30,540 an ability that Steve has not lost. 301 00:20:30,540 --> 00:20:34,180 But when we asked him to describe the picture, 302 00:20:34,180 --> 00:20:37,380 then he had a lot more difficulty there. 303 00:20:37,380 --> 00:20:41,580 Can you describe to me what you can see there? 304 00:20:43,100 --> 00:20:44,620 Um. 305 00:20:52,980 --> 00:20:55,460 What do you see here? 306 00:20:55,460 --> 00:20:57,020 Um... 307 00:21:02,220 --> 00:21:04,860 A cat. 308 00:21:04,860 --> 00:21:07,140 That's right. 309 00:21:07,140 --> 00:21:10,020 What's the cat doing? 310 00:21:13,100 --> 00:21:15,460 Can you see what's here? 311 00:21:17,020 --> 00:21:18,100 Yes. 312 00:21:19,540 --> 00:21:21,300 Um... 313 00:21:21,300 --> 00:21:23,100 Fish. 314 00:21:23,100 --> 00:21:24,100 That's right. 315 00:21:24,100 --> 00:21:26,900 Yeah, fish. That's right. 316 00:21:26,900 --> 00:21:30,340 And can you see what the cat's trying to do to the fish? 317 00:21:36,020 --> 00:21:37,180 I don't know. 318 00:21:40,060 --> 00:21:44,340 What was happening was the cat was putting his paw into the fish bowl 319 00:21:44,340 --> 00:21:48,660 to try and get the fish, and what Steve picked out on was the objects, 320 00:21:48,660 --> 00:21:51,220 the cat and the fish, but not the action. 321 00:21:54,900 --> 00:21:56,940 Book, the books. 322 00:21:56,940 --> 00:22:01,420 And likewise with the books, the books were falling off the shelf 323 00:22:01,420 --> 00:22:05,260 and they were about to hit the head of the sleeping man 324 00:22:05,260 --> 00:22:07,220 and what he said was "books". 325 00:22:08,300 --> 00:22:10,700 The head. Yes? 326 00:22:10,700 --> 00:22:14,500 'And then all of a sudden, you know, he had a little breakthrough.' 327 00:22:16,020 --> 00:22:20,100 Bouncing on the head. 328 00:22:20,500 --> 00:22:25,340 Yes, excellent. That's very good, thank you very much. 329 00:22:25,340 --> 00:22:28,580 Matching the problem Steve has in recalling words 330 00:22:28,580 --> 00:22:30,020 with his detailed scan 331 00:22:30,020 --> 00:22:32,500 has helped identify parts of the brain 332 00:22:32,500 --> 00:22:34,740 involved in language in all of us. 333 00:22:34,740 --> 00:22:40,220 This frontal area of our brain is associated with word retrieval. 334 00:22:40,220 --> 00:22:44,700 Everything that he said and described indicated that he understood 335 00:22:44,700 --> 00:22:49,140 what he was seeing but he just couldn't get the words out to describe it 336 00:22:49,140 --> 00:22:52,980 in a way that you'd need to if you had to talk to somebody else. 337 00:22:58,700 --> 00:23:02,540 By dividing the process of speech into such precise tasks, 338 00:23:02,540 --> 00:23:05,700 scientists like Cathy are now piecing together 339 00:23:05,700 --> 00:23:08,820 the first functional map of our language brain. 340 00:23:10,500 --> 00:23:15,580 From an area that is crucial in allowing us to hear and monitor the sound of our own voice... 341 00:23:17,580 --> 00:23:22,460 ..to an area that is critical for analysing the very structure of speech. 342 00:23:24,500 --> 00:23:27,100 There are all sorts of different components 343 00:23:27,100 --> 00:23:29,780 that we're putting together to make this model 344 00:23:29,780 --> 00:23:31,660 so it is like doing a jigsaw puzzle 345 00:23:31,660 --> 00:23:35,260 where you have lots of different pieces and as you work on it, 346 00:23:35,260 --> 00:23:37,500 you find new pieces but you also find, 347 00:23:37,500 --> 00:23:41,340 "Ah I've got a little cluster of pieces here that now fit together 348 00:23:41,340 --> 00:23:43,780 "into something very, very meaningful." 349 00:23:46,820 --> 00:23:50,900 It's very much work in progress but we are building the clearest picture 350 00:23:50,900 --> 00:23:54,660 ever of the areas of the brain that are involved in talking. 351 00:24:10,860 --> 00:24:12,500 But how does it all begin? 352 00:24:12,500 --> 00:24:14,780 COMPUTERISED VOICE: Hi, baby! 353 00:24:14,780 --> 00:24:19,220 When does a new brain switch on to language? 354 00:24:19,220 --> 00:24:21,140 COMPUTERISED VOICE: Hi, baby! 355 00:24:21,140 --> 00:24:22,900 Part of our research 356 00:24:22,900 --> 00:24:26,060 has to do with helping to understand the roots of, 357 00:24:26,060 --> 00:24:28,300 of language reception 358 00:24:28,300 --> 00:24:32,260 and how it is that babies begin to understand speech sounds. 359 00:24:32,260 --> 00:24:36,020 Professor William Fifer works with new born babies to investigate 360 00:24:36,020 --> 00:24:39,980 at what point they start to respond to the speech they hear around them. 361 00:24:45,020 --> 00:24:48,060 Early experiments used an adapted pacifier 362 00:24:48,060 --> 00:24:51,500 to measure how the baby's sucking responses changed 363 00:24:51,500 --> 00:24:54,060 when it was played different voices, 364 00:24:54,060 --> 00:24:58,420 but this technique relied on the baby being awake. 365 00:24:58,420 --> 00:25:01,180 He now measures new born brain activity 366 00:25:01,180 --> 00:25:03,180 using an array of electrodes 367 00:25:03,180 --> 00:25:06,420 and he can do that even when they're sleeping. 368 00:25:07,940 --> 00:25:11,100 So does this cute little baby have a name yet? 369 00:25:11,100 --> 00:25:14,460 Liliana. Liliana. All right. 370 00:25:14,460 --> 00:25:17,660 And what we're talking about today is how quickly she's picking up 371 00:25:17,660 --> 00:25:19,420 information about language already. 372 00:25:19,420 --> 00:25:21,860 So, as you know, she's been listening in the womb 373 00:25:21,860 --> 00:25:24,140 and we think that she's particularly prepared 374 00:25:24,140 --> 00:25:27,140 to want to listen to certain things, especially Mom's voice. 375 00:25:27,140 --> 00:25:28,860 Sorry, Dad, but you'll come soon. 376 00:25:28,860 --> 00:25:30,500 That's OK, don't worry. 377 00:25:30,500 --> 00:25:34,460 She'll stop listening to both of us soon enough, we know. 378 00:25:34,460 --> 00:25:39,420 Pfeifer wants to see how one-day-old Liliana responds to speech sounds 379 00:25:39,420 --> 00:25:41,780 generated by different voices. 380 00:25:41,780 --> 00:25:45,940 Her mother, a stranger or a computer saying the same thing. 381 00:25:45,940 --> 00:25:48,780 Which in this case is "Hello, Baby!" 382 00:25:50,380 --> 00:25:52,900 COMPUTERISED VOICE: Hi, baby! 383 00:25:52,900 --> 00:25:58,220 Hi, baby! Hi, baby! 384 00:25:58,220 --> 00:26:00,020 Hi, baby! 385 00:26:01,780 --> 00:26:04,260 Hi, baby! 386 00:26:04,260 --> 00:26:06,100 Hi, baby! 387 00:26:06,100 --> 00:26:07,660 Hi, baby! 388 00:26:07,660 --> 00:26:09,620 So when we do the experiment, 389 00:26:09,620 --> 00:26:14,540 you actually can watch baby's raw brain activity right on the screen. 390 00:26:14,540 --> 00:26:21,580 But later we process it to help us understand how much that baby is attending to any particular sound. 391 00:26:23,180 --> 00:26:27,380 Pfeifer found that Liliana is able to respond to her mother's voice 392 00:26:27,380 --> 00:26:29,420 in a way she doesn't to any other. 393 00:26:29,420 --> 00:26:30,580 Hi, baby! 394 00:26:30,580 --> 00:26:32,980 He believes the only explanation 395 00:26:32,980 --> 00:26:36,820 is that she'd been learning from her mother in the womb. 396 00:26:36,820 --> 00:26:40,500 Babies especially like to hear sounds that we call language 397 00:26:40,500 --> 00:26:43,340 and it's those speech sounds that they've heard 398 00:26:43,340 --> 00:26:45,220 that mother produced in utero 399 00:26:45,220 --> 00:26:47,980 that later they'll learn carry information 400 00:26:47,980 --> 00:26:50,420 that's going to be very important to them 401 00:26:50,420 --> 00:26:55,140 but when they're first hearing these sounds they don't have any real meaning 402 00:26:55,140 --> 00:26:57,340 but it's the quality of that voice, 403 00:26:57,340 --> 00:27:01,420 it's the number of times they hear it, it's the rhythm, the cadence 404 00:27:01,420 --> 00:27:06,100 that they're having some exposure to and we think that's what's affecting 405 00:27:06,100 --> 00:27:09,580 their early auditory system so, as soon as they're born, 406 00:27:09,580 --> 00:27:11,900 they cue in on that particular sound. 407 00:27:11,900 --> 00:27:14,940 With his work, Pfeifer is showing that we are tuned 408 00:27:14,940 --> 00:27:18,420 to the sounds of speech from the beginning of our lives. 409 00:27:19,940 --> 00:27:23,300 'My baby's room will be yellow.' 410 00:27:23,300 --> 00:27:25,940 BABY CRIES 411 00:27:25,940 --> 00:27:29,220 Don't sound very good. That was too loud. 412 00:27:29,220 --> 00:27:32,740 Yes, you didn't like her voice, no? 413 00:27:32,740 --> 00:27:36,300 But this unique ability doesn't last for long. 414 00:27:42,300 --> 00:27:44,580 THEY SPEAK IN NATIVE TONGUE 415 00:28:06,220 --> 00:28:09,780 What is it that makes learning language so easy as a child 416 00:28:09,780 --> 00:28:11,700 and so difficult as an adult? 417 00:28:13,020 --> 00:28:15,740 DOORBELL RINGS 418 00:28:16,780 --> 00:28:18,580 Hi, Chris. Hi. 419 00:28:18,580 --> 00:28:22,540 How are you? Como estas? Estoy bien. 420 00:28:22,540 --> 00:28:26,020 Christopher Taylor is helping to answer this question. 421 00:28:26,020 --> 00:28:30,660 Chris is autistic and lives in a care home in North Yorkshire. 422 00:28:30,660 --> 00:28:34,420 Despite being reliant on others for so many things in his life, 423 00:28:34,420 --> 00:28:36,500 in one way Christopher is unique. 424 00:28:36,500 --> 00:28:40,020 While he doesn't like talking to people very much, 425 00:28:40,020 --> 00:28:42,500 he can speak more than 20 languages 426 00:28:42,500 --> 00:28:46,020 including Welsh, Berber, Arabic and Icelandic. 427 00:28:46,020 --> 00:28:50,940 Espanol y que otra lengua? 428 00:28:50,940 --> 00:28:54,900 Que pasa aqui? How do you say zapatero in English? 429 00:28:54,900 --> 00:28:57,660 Shoe maker. Shoe maker, that's right. 430 00:28:58,860 --> 00:29:00,660 Language expert Gary Morgan 431 00:29:00,660 --> 00:29:04,380 has been working with Christopher for more than ten years. 432 00:29:04,380 --> 00:29:06,820 When I first used to talk to Christopher 433 00:29:06,820 --> 00:29:09,220 and watch the way he plays with language, 434 00:29:09,220 --> 00:29:11,660 the way he uses language, it was amazing. 435 00:29:11,660 --> 00:29:13,620 I'd never seen anything like it. 436 00:29:13,620 --> 00:29:17,380 OK, I've got some other papers here. Chinese. 437 00:29:17,380 --> 00:29:21,780 Yeah. This is in Serbian. Vesti. 438 00:29:21,780 --> 00:29:23,020 Serbian? 439 00:29:23,020 --> 00:29:26,860 CHRISTOPHER READS IN SERBIAN 440 00:29:26,860 --> 00:29:33,140 Learning new words is easy for Christopher. 441 00:29:33,140 --> 00:29:37,300 And today he's meeting Gerardo who's going to teach him words in Nahuatl, 442 00:29:37,300 --> 00:29:40,020 a language that's been spoken in Central Mexico 443 00:29:40,020 --> 00:29:41,580 since the seventh century. 444 00:29:41,580 --> 00:29:44,540 Y que necesito escribir? 445 00:29:44,540 --> 00:29:48,020 Si, escribo. Este es una palabra Nahuatl 446 00:29:48,020 --> 00:29:52,220 que se usan en espanol de Mexico y es papalotl. 447 00:29:52,220 --> 00:29:55,220 Mariposa. Mm-hmm. 448 00:29:55,220 --> 00:29:59,340 Christopher, when he hears a new word or hears about a new language 449 00:29:59,340 --> 00:30:01,980 he doesn't know, he really gets excited. 450 00:30:01,980 --> 00:30:05,980 It's what switches him on so, I think for Christopher, 451 00:30:05,980 --> 00:30:09,620 languages are like collecting rare butterflies. 452 00:30:09,620 --> 00:30:12,380 He likes to collect different languages, 453 00:30:12,380 --> 00:30:15,060 different words in different languages. 454 00:30:15,060 --> 00:30:18,020 Chris, you know these words that you learnt? Yes. 455 00:30:19,580 --> 00:30:22,780 Just tell, just tell them to me again really quickly. 456 00:30:22,780 --> 00:30:25,060 What was that one? Papalotl. 457 00:30:25,060 --> 00:30:27,860 How do you say it in, in, in English? 458 00:30:27,860 --> 00:30:29,820 Butterfly. 459 00:30:29,820 --> 00:30:31,660 Chilli. Chapulin. 460 00:30:31,660 --> 00:30:33,500 Cricket. Cocone. 461 00:30:33,500 --> 00:30:37,460 Children, pepper, flower. 462 00:30:37,460 --> 00:30:41,660 Oh, no lo conocemos, cochinilla, guajelotl. 463 00:30:41,660 --> 00:30:47,100 Turkey. Aqualotl? Tadpole. Perfecto. 464 00:30:47,100 --> 00:30:50,180 That's very quick. He does learn really quickly. 465 00:30:50,180 --> 00:30:53,460 Christopher memorised all of the words he was taught 466 00:30:53,460 --> 00:30:54,820 in just ten minutes. 467 00:30:54,820 --> 00:30:58,860 His incredible ability means he's continually adding new languages 468 00:30:58,860 --> 00:31:01,260 to the long list of those he's mastered. 469 00:31:03,060 --> 00:31:08,500 I think Christopher illustrates the ability to, to learn rules quickly, the patterns, 470 00:31:08,500 --> 00:31:11,260 scan through things quickly and see things, 471 00:31:11,260 --> 00:31:14,300 predict what's going to come next in the language. 472 00:31:14,300 --> 00:31:19,580 Christopher has this island of ability in a sea of disability 473 00:31:19,580 --> 00:31:24,860 so everything else is poor compared with one unique ability. 474 00:31:24,860 --> 00:31:28,580 The natural talent that shines through in Christopher 475 00:31:28,580 --> 00:31:31,260 seems to be revealing an extreme version 476 00:31:31,260 --> 00:31:34,180 of what might have once existed in all of us. 477 00:31:35,820 --> 00:31:37,540 An innate ability for language. 478 00:31:41,900 --> 00:31:44,900 Je m'appelle Frederica et je parle le francais, 479 00:31:44,900 --> 00:31:46,420 l'anglais et l'italien. 480 00:31:46,420 --> 00:31:48,940 HE SPEAKS IN NATIVE LANGUAGE 481 00:31:54,700 --> 00:31:57,580 THEY SPEAK IN NATIVE LANGUAGE 482 00:32:15,340 --> 00:32:20,900 D. D. Clever boy. Can you pick them up and tell Sylvie? 483 00:32:20,900 --> 00:32:22,780 S for Sylvie. 484 00:32:22,780 --> 00:32:24,540 What's that, darling? 485 00:32:24,540 --> 00:32:26,620 Monkey. Yeah. 486 00:32:26,620 --> 00:32:30,900 It was 50 years ago, the godfather of linguistics, Noam Chomsky, 487 00:32:30,900 --> 00:32:36,780 revolutionised the field by suggesting that the basis of our language ability was innate, 488 00:32:36,780 --> 00:32:38,460 built into each one of us. 489 00:32:38,460 --> 00:32:41,940 Now let's have a look at the book. 490 00:32:41,940 --> 00:32:46,540 He believes this explains why a child can learn language very easily 491 00:32:46,540 --> 00:32:48,540 despite its complex nature. 492 00:32:48,540 --> 00:32:52,340 It seems that the child almost reflexively picks up 493 00:32:52,340 --> 00:32:56,460 specific characteristics of the language in question, 494 00:32:56,460 --> 00:32:58,460 sound structure, pitch and so on, 495 00:32:58,460 --> 00:33:01,780 and picks out its words and somehow knows what they mean, 496 00:33:01,780 --> 00:33:04,460 and they mean very complex things. 497 00:33:04,460 --> 00:33:08,300 That's mummy. That's the mummy monkey, so who's this then? 498 00:33:08,300 --> 00:33:09,820 Daddy. Daddy. Daddy. 499 00:33:09,820 --> 00:33:12,940 But all this somehow gets picked up very quickly, 500 00:33:12,940 --> 00:33:16,900 which can only mean what it means in every aspect of biology. 501 00:33:16,900 --> 00:33:19,220 It's just the way the child is designed. 502 00:33:19,220 --> 00:33:21,980 Chomsky argues that our ability to talk 503 00:33:21,980 --> 00:33:25,460 is not just the result of our general intelligence 504 00:33:25,460 --> 00:33:30,140 but that we have an unconscious innate understanding about things 505 00:33:30,140 --> 00:33:34,380 like sentence structure, word order, meaning and sounds. 506 00:33:34,380 --> 00:33:36,020 Arlo, why don't you count them? 507 00:33:36,020 --> 00:33:39,420 But while we may have the blueprint for language, 508 00:33:39,420 --> 00:33:42,180 we need to be exposed to it early in life. 509 00:33:42,180 --> 00:33:44,740 If a child is brought up on stilts, let's say, 510 00:33:44,740 --> 00:33:48,260 and then they're taken off when it's two years old, 511 00:33:48,260 --> 00:33:51,420 it probably won't be able to acquire the capacity to walk 512 00:33:51,420 --> 00:33:53,020 because it's just too late. 513 00:33:53,020 --> 00:33:55,940 You know, something should have happened already. 514 00:33:55,940 --> 00:33:57,980 Same with binocular vision and so on, 515 00:33:57,980 --> 00:34:00,420 and probably the same is true of language. 516 00:34:00,420 --> 00:34:05,620 Chomsky's theories get to the heart of the fundamental question - 517 00:34:05,620 --> 00:34:09,140 is talking the result of nature or nurture? 518 00:34:13,300 --> 00:34:18,100 It's a problem that's notoriously difficult to unpick. 519 00:34:21,340 --> 00:34:25,740 One theoretical solution that has enticed scientists for many years 520 00:34:25,740 --> 00:34:29,260 would be to isolate a child and raise them in silence. 521 00:34:29,260 --> 00:34:32,540 Would the child develop language on their own? 522 00:34:32,540 --> 00:34:35,100 Because of the cruelty involved, 523 00:34:35,100 --> 00:34:39,420 scientists could never carry out this test on humans. 524 00:34:39,420 --> 00:34:43,900 It's become known as the Forbidden Experiment. 525 00:34:48,180 --> 00:34:50,620 The best scientists could hope for 526 00:34:50,620 --> 00:34:53,380 would be to find real live case studies, 527 00:34:53,380 --> 00:34:57,340 feral children, those brought up by animals in the wild. 528 00:34:58,620 --> 00:35:01,980 If children raised away from human contact could be found, 529 00:35:01,980 --> 00:35:05,780 how they communicated could finally answer the question 530 00:35:05,780 --> 00:35:08,580 of whether the ability to talk was innate. 531 00:35:15,300 --> 00:35:17,820 In 1991 the scientific community 532 00:35:17,820 --> 00:35:21,140 believed they'd stumbled across such a child. 533 00:35:21,140 --> 00:35:23,380 SHE BARKS 534 00:35:24,900 --> 00:35:29,460 Eight-year-old Oxana Malaya was found in the Ukraine 535 00:35:29,460 --> 00:35:32,100 and she'd been raised by dogs. 536 00:35:32,100 --> 00:35:33,740 They were surprised to find 537 00:35:33,740 --> 00:35:36,740 that while she behaved like the dogs she lived with, 538 00:35:36,740 --> 00:35:38,620 even barked to communicate... 539 00:35:38,620 --> 00:35:40,620 SHE BARKS 540 00:35:40,620 --> 00:35:44,220 She did speak some human words. 541 00:35:49,700 --> 00:35:54,220 The findings seemed to suggest that despite her circumstances, 542 00:35:54,220 --> 00:35:57,180 Oxana had an innate ability for language, 543 00:35:57,180 --> 00:36:00,700 one that was nurtured after she was found. 544 00:36:09,140 --> 00:36:11,260 SHE SPEAKS RUSSIAN 545 00:36:13,620 --> 00:36:17,220 But it later emerged that in her first three years 546 00:36:17,220 --> 00:36:21,020 she lived in a domestic setting and had had human interaction. 547 00:36:24,740 --> 00:36:28,340 It meant the scientists were no closer to establishing 548 00:36:28,340 --> 00:36:30,860 how much of our ability to talk is innate. 549 00:36:33,980 --> 00:36:38,860 The Forbidden Experiment has remained enticing but out of reach. 550 00:36:40,380 --> 00:36:41,940 Until now... 551 00:36:49,660 --> 00:36:51,900 One scientist has found a unique way 552 00:36:51,900 --> 00:36:54,300 of getting round the ethical problem. 553 00:37:01,820 --> 00:37:05,500 Ofer Tchernichovski is studying something linguists 554 00:37:05,500 --> 00:37:10,020 think might hold the key to whether speech and language is innate. 555 00:37:12,380 --> 00:37:19,940 The Forbidden Experiment is trying to see what speech and language 556 00:37:19,940 --> 00:37:24,180 human infants will develop without any exposure to spoken language. 557 00:37:24,180 --> 00:37:27,420 Of course, this cannot be done in humans 558 00:37:27,420 --> 00:37:30,580 but why shouldn't we do it with birds? 559 00:37:30,580 --> 00:37:34,780 Zebra finches are perfect subjects. 560 00:37:36,020 --> 00:37:38,420 While the females don't sing at all, 561 00:37:38,420 --> 00:37:42,260 the male chicks learn to sing by imitating their fathers, 562 00:37:42,260 --> 00:37:45,300 just like human children copy their parents. 563 00:37:55,380 --> 00:37:58,700 To recreate the conditions of the Forbidden Experiment, 564 00:37:58,700 --> 00:38:02,660 Ofer and his team isolated male chicks away from their fathers 565 00:38:02,660 --> 00:38:04,660 before they were taught to sing. 566 00:38:07,260 --> 00:38:11,180 Kept in isolation in specially sound-proofed cages, 567 00:38:11,180 --> 00:38:14,380 their song developed to no more than a croak. 568 00:38:24,100 --> 00:38:27,540 One of these isolated males was added to a group of females 569 00:38:27,540 --> 00:38:29,220 to create a new finch colony. 570 00:38:30,940 --> 00:38:33,300 Ofer hoped that the females would find 571 00:38:33,300 --> 00:38:35,380 the male's call attractive enough 572 00:38:35,380 --> 00:38:38,500 that they would mate and create the next generation. 573 00:38:41,260 --> 00:38:44,780 In the beginning, the first problem that we had is that the females 574 00:38:44,780 --> 00:38:47,180 were not very interested in the isolate male. 575 00:38:47,180 --> 00:38:50,700 When he start singing his isolate unstructured song, they say "yuk!" 576 00:38:50,700 --> 00:38:52,260 And they wouldn't go with him. 577 00:38:52,260 --> 00:38:56,780 But, we didn't let go and we kept going like that with a few tries 578 00:38:56,780 --> 00:39:00,220 and in the end one female became desperate enough 579 00:39:00,220 --> 00:39:02,900 and mated with the isolate founder, 580 00:39:02,900 --> 00:39:06,700 and then we had baby birds that imitated their father. 581 00:39:08,260 --> 00:39:11,540 The young birds instinctively copied their father's song 582 00:39:11,540 --> 00:39:14,260 despite its unstructured unattractive sound. 583 00:39:14,260 --> 00:39:16,060 And interestingly enough, 584 00:39:16,060 --> 00:39:19,660 in isolate song we see song that may be more prolonged. 585 00:39:19,660 --> 00:39:22,420 We can see syllable that go like "Waaaaah" like that 586 00:39:22,420 --> 00:39:24,300 instead of "choob-chab-choob." 587 00:39:24,300 --> 00:39:28,060 And this is something that they were still imitating 588 00:39:28,060 --> 00:39:32,740 so they were listening to that tutor that was singing abnormal song 589 00:39:32,740 --> 00:39:37,340 and they imitated his abnormal song, and so did their children. 590 00:39:40,740 --> 00:39:43,380 That in itself was not surprising. 591 00:39:43,380 --> 00:39:45,540 The sons were tutored by their father. 592 00:39:49,220 --> 00:39:53,860 But when the researchers compared song over the generations, 593 00:39:53,860 --> 00:39:56,700 they saw something extraordinary. 594 00:39:56,700 --> 00:39:56,700 Each new generation of birds didn't just imitate their father's song. 595 00:39:56,700 --> 00:40:02,500 They also improved it. 596 00:40:07,740 --> 00:40:09,380 By the fourth generation, 597 00:40:09,380 --> 00:40:13,500 the finches were singing a version of the song that normal birds sing, 598 00:40:13,500 --> 00:40:16,580 despite none of their colony having ever heard it. 599 00:40:22,140 --> 00:40:27,780 We were very pleased to see how quickly normal songs start emerging 600 00:40:27,780 --> 00:40:32,180 out of those very unstructured songs of the isolate 601 00:40:32,180 --> 00:40:37,900 and that individual bird has this ability to imitate and to improve, 602 00:40:37,900 --> 00:40:43,220 and improve at the level that he makes the song more species like 603 00:40:43,220 --> 00:40:45,740 and perhaps more effective. 604 00:40:45,740 --> 00:40:49,620 To adapt a song without being taught how to 605 00:40:49,620 --> 00:40:54,100 means the finches must have some innate ability. 606 00:40:54,100 --> 00:40:57,260 In the bird, seems to be somehow encoded 607 00:40:57,260 --> 00:40:59,780 so that when they imitate sounds, 608 00:40:59,780 --> 00:41:04,140 they'll shift them to where some kind of an internal image 609 00:41:04,140 --> 00:41:08,620 that they have of how a good zebra finch song should be like. 610 00:41:12,220 --> 00:41:15,580 Ofer's work suggests humans might also have 611 00:41:15,580 --> 00:41:17,780 this innate ability to talk. 612 00:41:20,180 --> 00:41:22,100 If we'd never heard speech, 613 00:41:22,100 --> 00:41:25,700 we'd develop a form of language within a few generations. 614 00:41:29,020 --> 00:41:31,860 After 50 years of research delving into our brains. 615 00:41:31,860 --> 00:41:34,420 Hello, Steve, can you hear me? 616 00:41:34,420 --> 00:41:36,940 Observing the birth of language in children. 617 00:41:36,940 --> 00:41:38,060 Hi, baby! 618 00:41:38,060 --> 00:41:41,620 Studying song birds and investigating anatomy... 619 00:41:43,340 --> 00:41:46,580 Scientists have come to the inescapable conclusion 620 00:41:46,580 --> 00:41:48,380 that we're built for speech, 621 00:41:50,940 --> 00:41:53,100 and that can only mean one thing. 622 00:41:53,100 --> 00:41:56,860 The origin of language must lie in our genes. 623 00:41:56,860 --> 00:42:00,020 It would be like looking for a needle in a haystack. 624 00:42:01,060 --> 00:42:03,620 THEY SPEAK IN NATIVE TONGUE 625 00:42:07,820 --> 00:42:12,700 Unpicking the genetic component of language is a huge challenge. 626 00:42:12,700 --> 00:42:16,620 From the thousands of genes that make up the human genome, 627 00:42:16,620 --> 00:42:20,340 it seemed almost impossible to identify a gene that has 628 00:42:20,340 --> 00:42:24,140 an influence on the multitude of functions behind speech. 629 00:42:26,100 --> 00:42:28,860 But 20 years ago Faraneh Vargha-Khadem, 630 00:42:28,860 --> 00:42:32,340 a neuroscientist from University College London, 631 00:42:32,340 --> 00:42:35,020 was introduced to a British family 632 00:42:35,020 --> 00:42:38,100 that had a very unusual speech defect. 633 00:42:38,100 --> 00:42:42,860 A family who would help solve the mystery of how we speak. 634 00:42:42,860 --> 00:42:48,780 When they are speaking they give the impression that they are deaf, 635 00:42:48,780 --> 00:42:50,980 but in fact they are hearing individuals 636 00:42:50,980 --> 00:42:53,420 but they give the impression that they are deaf 637 00:42:53,420 --> 00:42:55,260 because their speech is not clear 638 00:42:55,260 --> 00:42:57,900 so people have got difficulty understanding them. 639 00:42:59,700 --> 00:43:02,340 When our children was born, 640 00:43:02,340 --> 00:43:07,260 they found it very harder, or severe, for talking. 641 00:43:07,260 --> 00:43:13,660 But some people at school would take the mick out of me. 642 00:43:13,660 --> 00:43:17,020 Faraneh and her team spent several years testing the abilities 643 00:43:17,020 --> 00:43:19,740 of affected and unaffected members of the family 644 00:43:19,740 --> 00:43:22,140 so they could fully diagnose the condition. 645 00:43:22,140 --> 00:43:28,780 To protect their identity, they were named the KE family. 646 00:43:28,780 --> 00:43:32,660 When it comes to the KE family members who are affected 647 00:43:32,660 --> 00:43:35,580 with this speech and language disorder, 648 00:43:35,580 --> 00:43:38,420 they're very normal in every other way. 649 00:43:38,420 --> 00:43:42,900 They, of course, have a very specific speech and language problem 650 00:43:42,900 --> 00:43:47,460 but their thought processes and their communication abilities 651 00:43:47,460 --> 00:43:50,060 in other ways are perfectly normal. 652 00:43:50,060 --> 00:43:54,540 Hoping they would find a link between affected family members, 653 00:43:54,540 --> 00:43:56,780 a geneticist joined the team. 654 00:44:04,340 --> 00:44:08,660 The family at the time consisted of 30 members and 15 of the members 655 00:44:08,660 --> 00:44:13,100 had inherited a speech and language disorder from the grandmother. 656 00:44:13,100 --> 00:44:14,820 The first genetic study 657 00:44:14,820 --> 00:44:18,700 was able to locate the general area of the abnormality. 658 00:44:18,700 --> 00:44:24,580 It was on chromosome 7 and all the affected family members had it. 659 00:44:24,580 --> 00:44:29,260 Reflecting its obvious importance in their ability to talk, 660 00:44:29,260 --> 00:44:32,020 the research team called it Speech 1. 661 00:44:33,660 --> 00:44:36,820 Speech 1 was located on chromosome number 7 662 00:44:36,820 --> 00:44:39,740 but where exactly on chromosome number 7 663 00:44:39,740 --> 00:44:42,140 and which particular gene was it, 664 00:44:42,140 --> 00:44:44,940 was something that would have required 665 00:44:44,940 --> 00:44:47,780 several more years of investigations 666 00:44:47,780 --> 00:44:50,460 until the actual gene was detected. 667 00:44:50,460 --> 00:44:54,500 Then, out of the blue, a totally unconnected child 668 00:44:54,500 --> 00:44:59,380 with the same type of speech defect was brought in to see them. 669 00:44:59,380 --> 00:45:02,980 And when you think that in this big wide world, 670 00:45:02,980 --> 00:45:07,700 the same kind of problem presents itself in the same country 671 00:45:07,700 --> 00:45:11,140 in the same location but to the same person! 672 00:45:11,140 --> 00:45:15,980 So you really have to recognise that there is a huge element of luck 673 00:45:15,980 --> 00:45:20,180 that was involved in allowing this discovery to be made. 674 00:45:22,500 --> 00:45:25,140 Finally, they had the breakthrough they needed. 675 00:45:26,780 --> 00:45:28,660 Tests on the new child's DNA 676 00:45:28,660 --> 00:45:32,860 revealed that part of his chromosome 7 was clearly broken off. 677 00:45:34,740 --> 00:45:38,020 And to their amazement, Faraneh and her team 678 00:45:38,020 --> 00:45:42,660 found a mutation in the KE family's DNA, in exactly the same location. 679 00:45:42,660 --> 00:45:46,700 That finding suggests that many people with speech disorders 680 00:45:46,700 --> 00:45:48,300 could have a faulty gene. 681 00:45:49,580 --> 00:45:54,140 After years of searching, they'd finally found a gene, 682 00:45:54,140 --> 00:45:57,820 one that controls our unique ability to shape words. 683 00:45:57,820 --> 00:46:00,100 They called it FOXP2. 684 00:46:01,980 --> 00:46:04,820 FOXP2 seems to have a very important role 685 00:46:04,820 --> 00:46:07,340 on the control of this part of the face 686 00:46:07,340 --> 00:46:10,980 in order to be able to communicate the tiny movements 687 00:46:10,980 --> 00:46:14,740 that are required in order to sequence the speech sounds 688 00:46:14,740 --> 00:46:17,140 that results in our fluent speech. 689 00:46:17,140 --> 00:46:19,180 HE SPEAKS IN NATIVE TONGUE 690 00:46:19,180 --> 00:46:21,020 Your ability to talk. 691 00:46:21,020 --> 00:46:22,380 Your ability to talk. 692 00:46:22,380 --> 00:46:24,260 De la tua capacitad di parlare. 693 00:46:25,860 --> 00:46:27,860 The discovery of FOXP2 in humans 694 00:46:27,860 --> 00:46:31,740 marked a significant step forward for linguistic science. 695 00:46:33,420 --> 00:46:39,260 One family had unlocked some of the key secrets of speech in all of us. 696 00:46:39,260 --> 00:46:42,740 This kind of a connection between a single gene 697 00:46:42,740 --> 00:46:46,420 and a cognitive capacity, 698 00:46:46,420 --> 00:46:51,260 like speech and language, is quite remarkable to be able to demonstrate. 699 00:46:51,260 --> 00:46:54,140 It was later discovered that all vertebrates 700 00:46:54,140 --> 00:46:55,780 have a version of this gene 701 00:46:55,780 --> 00:46:58,420 but just two tiny changes in the human form 702 00:46:58,420 --> 00:47:02,980 has meant the difference between making noises and producing speech. 703 00:47:04,820 --> 00:47:09,900 It looks like this period where these changes occurred 704 00:47:09,900 --> 00:47:15,980 may have coincided with the time that modern humans had come about. 705 00:47:23,220 --> 00:47:27,020 The effect of these small changes in our genes was massive 706 00:47:27,020 --> 00:47:28,900 because, for the first time, 707 00:47:28,900 --> 00:47:32,780 we were able to manipulate our mouths to form complex sounds. 708 00:47:35,860 --> 00:47:39,940 Once we had the anatomy and physiology required for speech, 709 00:47:39,940 --> 00:47:41,620 we were primed to talk. 710 00:47:41,620 --> 00:47:46,380 Enriched their lives enormously. I don't know why I said it. 711 00:47:46,380 --> 00:47:49,020 To create everything that exists up till now. 712 00:47:51,020 --> 00:47:54,340 But when did we actually start to speak? 713 00:48:00,100 --> 00:48:03,660 The dawn of speech is one of the great puzzles of evolution 714 00:48:03,660 --> 00:48:07,260 because there's no fossil record of verbal communication 715 00:48:07,260 --> 00:48:09,460 for scientists to look back through. 716 00:48:11,700 --> 00:48:15,540 Mark Pagel is an evolutionary biologist 717 00:48:15,540 --> 00:48:19,220 on the hunt for the origins of language. 718 00:48:19,220 --> 00:48:21,060 There's a lot of ideas about this. 719 00:48:21,060 --> 00:48:22,620 They're all speculation 720 00:48:22,620 --> 00:48:25,740 but they rely on having the right genetics in place 721 00:48:25,740 --> 00:48:28,220 and having the right anatomy in place, 722 00:48:28,220 --> 00:48:30,780 and we know that by 200,000 years ago, 723 00:48:30,780 --> 00:48:33,420 both of those things were available. 724 00:48:36,700 --> 00:48:39,380 We may have been physically able to speak 725 00:48:39,380 --> 00:48:43,700 but linguists believe it's more likely that the birth of language 726 00:48:43,700 --> 00:48:47,300 coincided with the proliferation of man-made objects 727 00:48:47,300 --> 00:48:49,100 around 50,000 years ago. 728 00:48:49,100 --> 00:48:53,820 At this time there was an explosion in art and technology. 729 00:48:55,380 --> 00:48:59,140 We start to see, for the first time in the archaeological record, 730 00:48:59,140 --> 00:49:02,420 lots of different tool types, lots of different implements, 731 00:49:02,420 --> 00:49:04,740 lots of different huts that people lived in 732 00:49:04,740 --> 00:49:07,660 and so these things seemed to appear rather abruptly, 733 00:49:07,660 --> 00:49:10,780 suggesting that there was some change in the way we lived. 734 00:49:14,060 --> 00:49:17,780 Many scientists think that the birth of language 735 00:49:17,780 --> 00:49:20,220 instigated this cultural boom. 736 00:49:20,220 --> 00:49:21,980 But without a time machine, 737 00:49:21,980 --> 00:49:27,820 we may never know exactly when we started to talk. 738 00:49:27,820 --> 00:49:29,700 But what we do know is why. 739 00:49:31,020 --> 00:49:34,980 Mark Pagel believes that language probably developed 740 00:49:34,980 --> 00:49:40,100 as a practical way of defining roles and it formed a kind of social glue. 741 00:49:40,100 --> 00:49:44,340 We had something that no other animal does and that's what humans do 742 00:49:44,340 --> 00:49:46,060 is that they divide up the tasks. 743 00:49:46,060 --> 00:49:49,620 So, early in our evolution, somebody might have been gathering wood 744 00:49:49,620 --> 00:49:51,740 while another one was gathering water. 745 00:49:51,740 --> 00:49:53,900 Somebody might have been making the huts 746 00:49:53,900 --> 00:49:55,860 while someone else was making spears 747 00:49:55,860 --> 00:49:58,340 and, at the end of the day, all those tasks are shared 748 00:49:58,340 --> 00:50:01,900 and I think we can start to see that we need language to regulate that. 749 00:50:01,900 --> 00:50:05,060 We need language for people to be able to talk to each other 750 00:50:05,060 --> 00:50:08,100 about what they're doing and to be able to represent what 751 00:50:08,100 --> 00:50:11,900 their contribution is to the society as opposed to everyone else's. 752 00:50:14,100 --> 00:50:17,460 This motivation to clarify social tasks 753 00:50:17,460 --> 00:50:21,100 was the final push we needed to get us talking. 754 00:50:25,340 --> 00:50:29,540 How we adapted the first sounds we made into language 755 00:50:29,540 --> 00:50:32,820 is now being revealed for the first time. 756 00:50:32,820 --> 00:50:35,420 Because no-one designed English, say, 757 00:50:35,420 --> 00:50:38,780 no-one sat down and thought, "English would be useful 758 00:50:38,780 --> 00:50:40,620 "if it had relative clauses 759 00:50:40,620 --> 00:50:43,660 "or some other piece of grammar," it just happens, 760 00:50:43,660 --> 00:50:47,860 it happens by this blind unconscious process of transmission 761 00:50:47,860 --> 00:50:49,460 and, for the first time, 762 00:50:49,460 --> 00:50:52,540 we're able to see that happen in the laboratory. 763 00:50:54,220 --> 00:50:55,820 Renana. 764 00:50:59,460 --> 00:51:02,300 Nepilu. 765 00:51:02,300 --> 00:51:05,660 Can you remember this language? 766 00:51:05,660 --> 00:51:07,020 Lapiranana. 767 00:51:08,620 --> 00:51:12,900 It's one you've never seen or heard before. 768 00:51:12,900 --> 00:51:15,500 Kisafiki. 769 00:51:17,660 --> 00:51:19,900 It's been created by Simon Kirby 770 00:51:19,900 --> 00:51:23,700 and is made up of words that describe alien fruit. 771 00:51:25,580 --> 00:51:27,100 Neheki. 772 00:51:30,380 --> 00:51:34,620 Kirby and his team have devised an experiment where they can watch 773 00:51:34,620 --> 00:51:38,940 this language evolve over hundreds of years in just one afternoon. 774 00:51:38,940 --> 00:51:40,740 Nahuna. 775 00:51:40,740 --> 00:51:44,300 We start the experiment with this garbage language, 776 00:51:44,300 --> 00:51:45,860 this random language. 777 00:51:45,860 --> 00:51:49,260 In fact, to call it a language is, is in some sense misleading. 778 00:51:49,260 --> 00:51:50,620 It's not even a language. 779 00:51:52,180 --> 00:51:55,020 Nereki. 780 00:51:55,020 --> 00:51:58,300 In the beginning, the alien words are completely random. 781 00:51:58,300 --> 00:52:00,820 Pilu. With no common factors between them. 782 00:52:03,260 --> 00:52:06,740 Lapalu. Kirby's guinea pigs have to familiarise themselves 783 00:52:06,740 --> 00:52:11,300 with the new words before being tested to see what they remember. 784 00:52:18,020 --> 00:52:20,340 But there's a twist. 785 00:52:20,340 --> 00:52:23,500 Some of the fruit are familiar 786 00:52:23,500 --> 00:52:26,860 but half of them have never been seen 787 00:52:26,860 --> 00:52:30,700 and the volunteers couldn't possibly know the words that describe them. 788 00:52:33,380 --> 00:52:37,500 Something most of them don't even notice. 789 00:52:39,460 --> 00:52:41,460 Kinamula. 790 00:52:46,740 --> 00:52:48,260 Neki. 791 00:52:48,260 --> 00:52:54,660 The early participants do very badly in the test because the language is completely random and unstructured. 792 00:52:54,660 --> 00:53:00,540 Unfortunately you didn't get any of them right. Oh, my God! Sorry. 793 00:53:00,540 --> 00:53:05,100 It was really hard. I just couldn't seem to find any patterns. 794 00:53:05,100 --> 00:53:08,660 I thought I'd be able to see a pattern with maybe the colours 795 00:53:08,660 --> 00:53:10,420 and I put in key or ka for yellow 796 00:53:10,420 --> 00:53:12,940 and then maybe I thought lu could mean two. 797 00:53:12,940 --> 00:53:16,620 It was a bit of a shock to hear I hadn't seen half of the language. 798 00:53:16,620 --> 00:53:19,860 I just felt a bit cheated, really! I didn't have a chance. 799 00:53:23,100 --> 00:53:24,740 Napilu. 800 00:53:24,740 --> 00:53:26,740 But the experiment doesn't end there. 801 00:53:26,740 --> 00:53:29,700 All the words produced by the first candidate 802 00:53:29,700 --> 00:53:33,260 are used to create the language for the next person in the experiment, 803 00:53:33,260 --> 00:53:37,820 including the words they invented for the unfamiliar fruit. 804 00:53:37,820 --> 00:53:41,980 Each of these learners thinks that they're giving us back 805 00:53:41,980 --> 00:53:45,980 the same thing that we trained them on, as best they can, 806 00:53:45,980 --> 00:53:48,900 but in fact each of them, unconsciously, 807 00:53:48,900 --> 00:53:53,380 is changing that language, changing it piece by piece over time. 808 00:53:56,300 --> 00:53:57,980 Nahuna. 809 00:53:57,980 --> 00:54:00,660 As it's passed through generations of users, 810 00:54:00,660 --> 00:54:03,700 the language slowly turns from a random chaotic one 811 00:54:03,700 --> 00:54:05,940 to one of structure with combinations 812 00:54:05,940 --> 00:54:07,780 that can be easily remembered. 813 00:54:13,220 --> 00:54:17,140 By the ninth generation, the words have been divided into parts 814 00:54:17,140 --> 00:54:19,340 and each of these has a different meaning. 815 00:54:20,980 --> 00:54:24,540 N, L and R describe the different colours. 816 00:54:24,540 --> 00:54:26,940 The middle of the word describes the number, 817 00:54:26,940 --> 00:54:29,700 and the end of the word like "plo" or "pilu" 818 00:54:29,700 --> 00:54:31,500 describes the type of fruit. 819 00:54:33,500 --> 00:54:36,940 This allows the parts to be joined together in different ways 820 00:54:36,940 --> 00:54:40,100 to describe fruit the volunteers have never seen before, 821 00:54:40,100 --> 00:54:42,300 a crucial feature of human language. 822 00:54:42,300 --> 00:54:44,900 I think lahoki will be this one. 823 00:54:44,900 --> 00:54:47,540 Got all five out of five right. Oh, good, excellent. 824 00:54:47,540 --> 00:54:51,060 The really striking thing that we've discovered 825 00:54:51,060 --> 00:54:53,700 is that when we run these experiments 826 00:54:53,700 --> 00:54:56,980 what we get at the end looks like a real language, 827 00:54:56,980 --> 00:55:00,660 not one of the languages that we see in the world today 828 00:55:00,660 --> 00:55:04,060 but it looks like a language in that it's got those 829 00:55:04,060 --> 00:55:08,420 essential defining characteristics that we see in language. 830 00:55:08,420 --> 00:55:10,340 In other words 831 00:55:10,340 --> 00:55:15,340 these languages that emerge are made up of parts that can be recombined. 832 00:55:15,340 --> 00:55:19,980 This happens gradually, it happens without anyone knowing it 833 00:55:19,980 --> 00:55:25,060 and yet it is the exact feature of language that makes it unique, 834 00:55:25,060 --> 00:55:30,300 makes language different from everything else we see in nature. 835 00:55:30,300 --> 00:55:32,700 Renana. 836 00:55:35,580 --> 00:55:39,700 I'll do it in Catalan. 837 00:55:41,540 --> 00:55:46,060 It's this ability we have to combine sounds to make new meanings 838 00:55:46,060 --> 00:55:48,940 that means how we talk will continue to change 839 00:55:48,940 --> 00:55:51,100 from generation to generation. 840 00:55:51,100 --> 00:55:53,020 That's mummy. 841 00:55:53,020 --> 00:55:57,380 And the fact that it's a living entity makes it so fascinating 842 00:55:57,380 --> 00:56:01,140 but so evasive for scientists. 843 00:56:01,140 --> 00:56:04,380 We have come a long way but, at the same time, 844 00:56:04,380 --> 00:56:07,260 it's really a window that has opened. 845 00:56:07,260 --> 00:56:09,940 We are only beginning to get a glimpse 846 00:56:09,940 --> 00:56:12,980 of how complicated the whole process is. 847 00:56:12,980 --> 00:56:15,540 Researchers are beginning to understand 848 00:56:15,540 --> 00:56:18,460 that, beyond our innate ability and our upbringing, 849 00:56:18,460 --> 00:56:21,540 there are more factors that influence why we talk. 850 00:56:21,540 --> 00:56:25,700 One area of growing consensus in understanding the biology of language 851 00:56:25,700 --> 00:56:29,780 is the realisation that language actually is not one monolithic thing 852 00:56:29,780 --> 00:56:33,100 but instead is made up of multiple different components. 853 00:56:33,100 --> 00:56:35,140 Come on, ?4 a pound, bananas! 854 00:56:35,140 --> 00:56:37,020 BABY BABBLES 855 00:56:37,020 --> 00:56:39,900 HE SPEAKS SERBIAN 856 00:56:39,900 --> 00:56:43,660 We talk about the innate abilities and the learnt abilities, 857 00:56:43,660 --> 00:56:47,860 what we inherit and what we acquire, what we're given by our culture, 858 00:56:47,860 --> 00:56:51,380 our society, our parents, and what we get from our genes. 859 00:56:51,380 --> 00:56:53,940 Chapulin. Cricket. 860 00:56:53,940 --> 00:56:55,580 Cocone. Children. 861 00:56:55,580 --> 00:56:59,900 I think both of those abilities, both of those sets of things 862 00:56:59,900 --> 00:57:02,980 need to be in place for you to have the full, 863 00:57:02,980 --> 00:57:06,060 fully-fledged abilities with language. 864 00:57:06,060 --> 00:57:09,500 And being able to identify all these components 865 00:57:09,500 --> 00:57:14,420 and understand how they fit together may tell us what it is to be human. 866 00:57:14,420 --> 00:57:15,860 A cat. 867 00:57:17,140 --> 00:57:22,380 And somehow out of this complicated soup of interactions between people, 868 00:57:22,380 --> 00:57:26,860 language emerges, and that's really what we need to understand 869 00:57:26,860 --> 00:57:30,380 if we are to understand where language comes from 870 00:57:30,380 --> 00:57:32,140 and why it is the way it is, 871 00:57:32,140 --> 00:57:35,020 and it's a really difficult problem. 872 00:57:35,020 --> 00:57:38,020 But answering the problem is important because 873 00:57:38,020 --> 00:57:40,300 it defines us, it defines who we are. 874 00:57:40,300 --> 00:57:42,340 Nehilu. 875 00:58:01,980 --> 00:58:04,980 Subtitles by Red Bee Media Ltd 876 00:58:04,980 --> 00:58:07,980 E-mail subtitling@bbc.co.uk