TED2011
Deb Roy: The birth of a word
Deb Roy：一個字詞的誕生
Filmed:
Readability: 4.2
2,809,941 views
MIT研究員Deb Roy想了解他襁褓中的兒子如何學習語言，因此在家裡裝設了攝影機，以便了解他兒子在生命中的每一個時刻（包含例外），然後分析了九萬個小時的家庭影片，觀察「gaaaa」慢慢地轉變為「water」的過程。在驚人且豐富的資料研究之下，富含深遠意義，幫助了解我們是如何學習。
Deb Roy - Cognitive scientist
Deb Roy studies how children learn language, and designs machines that learn to communicate in human-like ways. On sabbatical from MIT Media Lab, he's working with the AI company Bluefin Labs. Full bio
Deb Roy studies how children learn language, and designs machines that learn to communicate in human-like ways. On sabbatical from MIT Media Lab, he's working with the AI company Bluefin Labs. Full bio
Double-click the English transcript below to play the video.
00:15
Imagine if you could record your life --
0
0
4000
想一想，如果你能記錄下你的生命－
00:19
everything you said, everything you did,
1
4000
3000
你的一言、你的一行
00:22
available in a perfect memory store at your fingertips,
2
7000
3000
指頭點按一下，即可從完美的記憶體取得
00:25
so you could go back
3
10000
2000
那麽你便能回顧
00:27
and find memorable moments and relive them,
4
12000
3000
重溫值得回憶的時刻
00:30
or sift through traces of time
5
15000
3000
或是在過去的時光中篩選
00:33
and discover patterns in your own life
6
18000
2000
並發現自己的生命中
00:35
that previously had gone undiscovered.
7
20000
3000
先前沒有注意到的模式
00:38
Well that's exactly the journey
8
23000
2000
那就是我的家庭
00:40
that my family began
9
25000
2000
在過去五年半以來
00:42
five and a half years ago.
10
27000
2000
經過的歷程
00:44
This is my wife and collaborator, Rupal.
11
29000
3000
這是我太太也是同事Rupal
00:47
And on this day, at this moment,
12
32000
2000
就是在那一天的那一刻
00:49
we walked into the house with our first child,
13
34000
2000
我們帶著第一個孩子
00:51
our beautiful baby boy.
14
36000
2000
漂亮的兒子進入家門
00:53
And we walked into a house
15
38000
3000
我們走進的房子是一棟
00:56
with a very special home video recording system.
16
41000
4000
裝配了非常特別的錄影系統的房子
01:07
(Video) Man: Okay.
17
52000
2000
(影片) 男聲：好了
01:10
Deb Roy: This moment
18
55000
1000
Deb Roy：這個時刻以及
01:11
and thousands of other moments special for us
19
56000
3000
其他無數個對我們而言是特別的時刻
01:14
were captured in our home
20
59000
2000
都在我們家被記錄下來了
01:16
because in every room in the house,
21
61000
2000
因爲在房裡的每個房間
01:18
if you looked up, you'd see a camera and a microphone,
22
63000
3000
如果往上看就會看到攝影機和麥克風
01:21
and if you looked down,
23
66000
2000
如果往下看
01:23
you'd get this bird's-eye view of the room.
24
68000
2000
就能鳥瞰這房間
01:25
Here's our living room,
25
70000
3000
這是我們的客廳
01:28
the baby bedroom,
26
73000
3000
這是嬰兒房
01:31
kitchen, dining room
27
76000
2000
廚房、飯廳
01:33
and the rest of the house.
28
78000
2000
還有房子的其他地方
01:35
And all of these fed into a disc array
29
80000
3000
這些都儲存到一組設計來
01:38
that was designed for a continuous capture.
30
83000
3000
攫取連續影音的硬碟中
01:41
So here we are flying through a day in our home
31
86000
3000
現在快速看一下我們家的一天
01:44
as we move from sunlit morning
32
89000
3000
從太陽升起的早晨
01:47
through incandescent evening
33
92000
2000
到點亮燈火的夜晚
01:49
and, finally, lights out for the day.
34
94000
3000
到最後熄燈就寢
01:53
Over the course of three years,
35
98000
3000
這整整三年裡
01:56
we recorded eight to 10 hours a day,
36
101000
2000
我們每天紀錄
01:58
amassing roughly a quarter-million hours
37
103000
3000
8到10個小時
02:01
of multi-track audio and video.
38
106000
3000
積聚了大約25萬小時的影音
02:04
So you're looking at a piece of what is by far
39
109000
2000
大家現在看到的是有史以來
02:06
the largest home video collection ever made.
40
111000
2000
最長的家庭影集的一小部分
02:08
(Laughter)
41
113000
3000
(笑聲)
02:11
And what this data represents
42
116000
2000
這些資料所代表的
02:13
for our family at a personal level,
43
118000
4000
對我的家庭而言，在個人的層面上
02:17
the impact has already been immense,
44
122000
2000
已有巨大的影響
02:19
and we're still learning its value.
45
124000
3000
我們仍須進一步了解其中的意義
02:22
Countless moments
46
127000
2000
有無數的時刻
02:24
of unsolicited natural moments, not posed moments,
47
129000
3000
無預期、不造作的自然時刻
02:27
are captured there,
48
132000
2000
都被記錄了起來
02:29
and we're starting to learn how to discover them and find them.
49
134000
3000
我們還在學習如何發現找到那些時刻
02:32
But there's also a scientific reason that drove this project,
50
137000
3000
不過這個專案也有其科學上的目的
02:35
which was to use this natural longitudinal data
51
140000
4000
爲的是要利用這個縱向紀錄的資料
02:39
to understand the process
52
144000
2000
來了解一個小孩
02:41
of how a child learns language --
53
146000
2000
如何學習語言－
02:43
that child being my son.
54
148000
2000
那個小孩就是我的兒子
02:45
And so with many privacy provisions put in place
55
150000
4000
在設置隱私保護的條件下
02:49
to protect everyone who was recorded in the data,
56
154000
3000
讓每個在影片中的人都受到保護
02:52
we made elements of the data available
57
157000
3000
我們讓我信任的MIT
02:55
to my trusted research team at MIT
58
160000
3000
研究團隊取用這些資料
02:58
so we could start teasing apart patterns
59
163000
3000
於是我們可以開始從這巨大的
03:01
in this massive data set,
60
166000
3000
資料集裡拆解出其中的模式
03:04
trying to understand the influence of social environments
61
169000
3000
試圖了解社交環境
03:07
on language acquisition.
62
172000
2000
對語言習得有哪些影響
03:09
So we're looking here
63
174000
2000
我們看到這裡
03:11
at one of the first things we started to do.
64
176000
2000
這是我們首先進行的
03:13
This is my wife and I cooking breakfast in the kitchen,
65
178000
4000
這是太太和我在廚房做早餐
03:17
and as we move through space and through time,
66
182000
3000
當我們在時空中移動
03:20
a very everyday pattern of life in the kitchen.
67
185000
3000
會出現一條在廚房裡的日常生活軌跡
03:23
In order to convert
68
188000
2000
爲了轉換
03:25
this opaque, 90,000 hours of video
69
190000
3000
這九萬小時的影像
03:28
into something that we could start to see,
70
193000
2000
成爲能辨識的東西
03:30
we use motion analysis to pull out,
71
195000
2000
我們利用動作分析
03:32
as we move through space and through time,
72
197000
2000
汲取我們的移動軌跡
03:34
what we call space-time worms.
73
199000
3000
我們稱之爲「時空蟲」
03:37
And this has become part of our toolkit
74
202000
3000
這是我們的工具之一
03:40
for being able to look and see
75
205000
3000
用來查看
03:43
where the activities are in the data,
76
208000
2000
資料中所發生的活動
03:45
and with it, trace the pattern of, in particular,
77
210000
3000
再利用此法去追蹤，特別是
03:48
where my son moved throughout the home,
78
213000
2000
我兒子在家中活動的軌跡
03:50
so that we could focus our transcription efforts,
79
215000
3000
好讓我們能聚焦在解讀
03:53
all of the speech environment around my son --
80
218000
3000
有關我兒子學說話的環境－
03:56
all of the words that he heard from myself, my wife, our nanny,
81
221000
3000
所有他聽到的我、我太太和保姆說的字詞
03:59
and over time, the words he began to produce.
82
224000
3000
然後經過長時間他開始說那樣的字詞
04:02
So with that technology and that data
83
227000
3000
利用那樣的科技和資料
04:05
and the ability to, with machine assistance,
84
230000
2000
再利用機器輔助
04:07
transcribe speech,
85
232000
2000
便能轉譯他說出的言語
04:09
we've now transcribed
86
234000
2000
我們現在已經轉譯完成
04:11
well over seven million words of our home transcripts.
87
236000
3000
足足超過700萬個家中言談的字詞
04:14
And with that, let me take you now
88
239000
2000
我們利用這樣的轉譯來
04:16
for a first tour into the data.
89
241000
3000
瀏覽一下這些資料
04:19
So you've all, I'm sure,
90
244000
2000
我相信各位一定
04:21
seen time-lapse videos
91
246000
2000
都看過時間推移影片
04:23
where a flower will blossom as you accelerate time.
92
248000
3000
加快時間推移就可以看到花朵綻放
04:26
I'd like you to now experience
93
251000
2000
我現在想讓各位體驗一下
04:28
the blossoming of a speech form.
94
253000
2000
言語的花朵是怎麽綻放的
04:30
My son, soon after his first birthday,
95
255000
2000
我兒子過第一個生日後不久
04:32
would say "gaga" to mean water.
96
257000
3000
會說「gaga」來表示「水」
04:35
And over the course of the next half-year,
97
260000
3000
再接下來的半年時間裡
04:38
he slowly learned to approximate
98
263000
2000
他慢慢學會說出接近
04:40
the proper adult form, "water."
99
265000
3000
成年人說的正確的「water」
04:43
So we're going to cruise through half a year
100
268000
2000
我們現在要用40秒
04:45
in about 40 seconds.
101
270000
2000
快步瀏覽半年的歷程
04:47
No video here,
102
272000
2000
這裡不播放影片
04:49
so you can focus on the sound, the acoustics,
103
274000
3000
這樣各位便能聚焦在聲音的
04:52
of a new kind of trajectory:
104
277000
2000
這種新的軌跡變化
04:54
gaga to water.
105
279000
2000
從gaga到水
04:56
(Audio) Baby: Gagagagagaga
106
281000
12000
(語音) 嬰兒：Gagagagagaga
05:08
Gaga gaga gaga
107
293000
4000
Gaga gaga gaga
05:12
guga guga guga
108
297000
5000
guga guga guga
05:17
wada gaga gaga guga gaga
109
302000
5000
wada gaga gaga guga gaga
05:22
wader guga guga
110
307000
4000
wader guga guga
05:26
water water water
111
311000
3000
water water water
05:29
water water water
112
314000
6000
water water water
05:35
water water
113
320000
4000
water water
05:39
water.
114
324000
2000
water
05:41
DR: He sure nailed it, didn't he.
115
326000
2000
DR: 他的確說中了吧
05:43
(Applause)
116
328000
7000
(掌聲)
05:50
So he didn't just learn water.
117
335000
2000
他不僅學到「水」這個字詞
05:52
Over the course of the 24 months,
118
337000
2000
在24個月的歷程裡
05:54
the first two years that we really focused on,
119
339000
3000
頭兩年我們專注在這事上
05:57
this is a map of every word he learned in chronological order.
120
342000
4000
這裡有張圖列出他學到的所有字詞的時序
06:01
And because we have full transcripts,
121
346000
3000
由於我們有完整的轉譯
06:04
we've identified each of the 503 words
122
349000
2000
我們辨識出他在第二個生日前學到的
06:06
that he learned to produce by his second birthday.
123
351000
2000
這503個字詞的每一個
06:08
He was an early talker.
124
353000
2000
他算是早說話的
06:10
And so we started to analyze why.
125
355000
3000
所以我們開始分析原因
06:13
Why were certain words born before others?
126
358000
3000
爲什麽有些字詞來得早？
06:16
This is one of the first results
127
361000
2000
這是最早發現的結果之一
06:18
that came out of our study a little over a year ago
128
363000
2000
大約一年前的研究結果
06:20
that really surprised us.
129
365000
2000
真的很讓我們吃驚
06:22
The way to interpret this apparently simple graph
130
367000
3000
這張圖看來簡單，但解讀起來
06:25
is, on the vertical is an indication
131
370000
2000
在垂直線上有一顯示
06:27
of how complex caregiver utterances are
132
372000
3000
從言語長度來看，照顧者的話語
06:30
based on the length of utterances.
133
375000
2000
是很複雜的
06:32
And the [horizontal] axis is time.
134
377000
3000
垂直軸線表示時間
06:35
And all of the data,
135
380000
2000
我們將所有的資料
06:37
we aligned based on the following idea:
136
382000
3000
根據下述的想法排列：
06:40
Every time my son would learn a word,
137
385000
3000
每當發現我兒子就要學會一個字詞
06:43
we would trace back and look at all of the language he heard
138
388000
3000
我們就回溯查看他先前聽到出現
06:46
that contained that word.
139
391000
2000
那個字詞的所有言語裡
06:48
And we would plot the relative length of the utterances.
140
393000
4000
我們就繪製出那個比較長的言語
06:52
And what we found was this curious phenomena,
141
397000
3000
結果我們發現這麽一個奇特的現象
06:55
that caregiver speech would systematically dip to a minimum,
142
400000
3000
照顧者都會有系統地把字詞減降到最少
06:58
making language as simple as possible,
143
403000
3000
讓語言儘量變得簡單
07:01
and then slowly ascend back up in complexity.
144
406000
3000
然後又逐漸升回到複雜
07:04
And the amazing thing was
145
409000
2000
令人訝異的是
07:06
that bounce, that dip,
146
411000
2000
那個升、那個降
07:08
lined up almost precisely
147
413000
2000
幾乎與每個字詞
07:10
with when each word was born --
148
415000
2000
誕生的時間恰恰吻合－
07:12
word after word, systematically.
149
417000
2000
一詞接一詞，很有系統
07:14
So it appears that all three primary caregivers --
150
419000
2000
因此看來所有三個主要的照顧者－
07:16
myself, my wife and our nanny --
151
421000
3000
我自己、太太和我們的保姆－
07:19
were systematically and, I would think, subconsciously
152
424000
3000
都有系統地，而且我認為是下意識地
07:22
restructuring our language
153
427000
2000
重新建構我們的語言
07:24
to meet him at the birth of a word
154
429000
3000
好迎接一個字詞的誕生
07:27
and bring him gently into more complex language.
155
432000
4000
然後讓我兒子慢慢學習更複雜的語言
07:31
And the implications of this -- there are many,
156
436000
2000
這之中蘊含了許多意義
07:33
but one I just want to point out,
157
438000
2000
但我現在想指出的一點是
07:35
is that there must be amazing feedback loops.
158
440000
3000
這過程中必有令人驚異的反應循環
07:38
Of course, my son is learning
159
443000
2000
當然，我的兒子正在
07:40
from his linguistic environment,
160
445000
2000
從他的語言環境中學習
07:42
but the environment is learning from him.
161
447000
3000
但他所處的環境也會跟著有調整
07:45
That environment, people, are in these tight feedback loops
162
450000
3000
環境和人也都在這緊密的反應循環裡
07:48
and creating a kind of scaffolding
163
453000
2000
彼此建立起某種
07:50
that has not been noticed until now.
164
455000
3000
以往沒有注意到的橋梁
07:54
But that's looking at the speech context.
165
459000
2000
不過，那是從言語情境來看
07:56
What about the visual context?
166
461000
2000
若從視覺情境來看又是如何呢？
07:58
We're not looking at --
167
463000
2000
現在看到的還不是－
08:00
think of this as a dollhouse cutaway of our house.
168
465000
2000
想像這是我家剪下來的玩具屋
08:02
We've taken those circular fish-eye lens cameras,
169
467000
3000
我們使用環狀連結的魚眼攝影機
08:05
and we've done some optical correction,
170
470000
2000
然後再做一些光學修正
08:07
and then we can bring it into three-dimensional life.
171
472000
4000
因此我們可以做成3D影像
08:11
So welcome to my home.
172
476000
2000
歡迎來到我家
08:13
This is a moment,
173
478000
2000
這是其中一個時刻
08:15
one moment captured across multiple cameras.
174
480000
3000
透過多重攝影機攫取的一個時刻
08:18
The reason we did this is to create the ultimate memory machine,
175
483000
3000
這樣做是要創造出最高端的記憶機器
08:21
where you can go back and interactively fly around
176
486000
3000
可在其中以互動方式前後快速地搜尋
08:24
and then breathe video-life into this system.
177
489000
3000
從而爲此系統的影像注入生息
08:27
What I'm going to do
178
492000
2000
我現在要讓各位
08:29
is give you an accelerated view of 30 minutes,
179
494000
3000
看的是壓縮30分鐘的加快影像
08:32
again, of just life in the living room.
180
497000
2000
這次也只在客廳裡
08:34
That's me and my son on the floor.
181
499000
3000
那是我和我兒子在地板上
08:37
And there's video analytics
182
502000
2000
加上了影片分析
08:39
that are tracking our movements.
183
504000
2000
追蹤我們的動作
08:41
My son is leaving red ink. I am leaving green ink.
184
506000
3000
我兒子留下紅色軌迹，我留下綠色軌迹
08:44
We're now on the couch,
185
509000
2000
我們現在在沙發上
08:46
looking out through the window at cars passing by.
186
511000
3000
看著窗外駛過的汽車
08:49
And finally, my son playing in a walking toy by himself.
187
514000
3000
最後我兒子獨自玩會動的玩具
08:52
Now we freeze the action, 30 minutes,
188
517000
3000
我們在此停格，這段有30分鐘
08:55
we turn time into the vertical axis,
189
520000
2000
我們把時間放到垂直軸上
08:57
and we open up for a view
190
522000
2000
然後從中來看一下
08:59
of these interaction traces we've just left behind.
191
524000
3000
剛才留下來的互動軌迹
09:02
And we see these amazing structures --
192
527000
3000
我們看到這個令人訝異的結構－
09:05
these little knots of two colors of thread
193
530000
3000
這些兩種顔色的小結點
09:08
we call "social hot spots."
194
533000
2000
我們稱之爲社交熱點
09:10
The spiral thread
195
535000
2000
這個蜿蜒交纏的點串
09:12
we call a "solo hot spot."
196
537000
2000
我們稱之爲單一熱點
09:14
And we think that these affect the way language is learned.
197
539000
3000
我們認爲這些熱點對語言學習有影響
09:17
What we'd like to do
198
542000
2000
我們想要做的是
09:19
is start understanding
199
544000
2000
開始理解
09:21
the interaction between these patterns
200
546000
2000
這些模式與我兒子接觸的
09:23
and the language that my son is exposed to
201
548000
2000
語言之間的互動關係
09:25
to see if we can predict
202
550000
2000
看看是否能夠預測
09:27
how the structure of when words are heard
203
552000
2000
聽到字詞時的結構
09:29
affects when they're learned --
204
554000
2000
如何影響到字詞的學習－
09:31
so in other words, the relationship
205
556000
2000
換句話說就是
09:33
between words and what they're about in the world.
206
558000
4000
字詞與現實世界之間的關係
09:37
So here's how we're approaching this.
207
562000
2000
這就是我們的解讀方法
09:39
In this video,
208
564000
2000
在這個影片裡
09:41
again, my son is being traced out.
209
566000
2000
同樣是追蹤我的兒子
09:43
He's leaving red ink behind.
210
568000
2000
他留下紅色的軌迹
09:45
And there's our nanny by the door.
211
570000
2000
門旁的是我們的保姆
09:47
(Video) Nanny: You want water? (Baby: Aaaa.)
212
572000
3000
(影片) 保姆：你要喝水？(嬰孩：Aaaa.)
09:50
Nanny: All right. (Baby: Aaaa.)
213
575000
3000
保姆：好的 (嬰孩：Aaaa.)
09:53
DR: She offers water,
214
578000
2000
DR：她問他要不要水
09:55
and off go the two worms
215
580000
2000
然後兩條時空蟲
09:57
over to the kitchen to get water.
216
582000
2000
開始蠕動到廚房拿水
09:59
And what we've done is use the word "water"
217
584000
2000
我們用來標示那個時刻
10:01
to tag that moment, that bit of activity.
218
586000
2000
那個活動的字詞是「water」
10:03
And now we take the power of data
219
588000
2000
我們現在有龐大的資料
10:05
and take every time my son
220
590000
3000
從中汲取我兒子何時聽到
10:08
ever heard the word water
221
593000
2000
「水」這個字以及
10:10
and the context he saw it in,
222
595000
2000
他看到水的情境
10:12
and we use it to penetrate through the video
223
597000
3000
我們利用來這些透析整個影片
10:15
and find every activity trace
224
600000
3000
找出每一個與水相關
10:18
that co-occurred with an instance of water.
225
603000
3000
同時發生的活動的軌迹
10:21
And what this data leaves in its wake
226
606000
2000
這些資料留下了
10:23
is a landscape.
227
608000
2000
一幅風景
10:25
We call these wordscapes.
228
610000
2000
我們稱之爲「字詞風景」
10:27
This is the wordscape for the word water,
229
612000
2000
這就是water的「字詞風景」
10:29
and you can see most of the action is in the kitchen.
230
614000
2000
各位可以看到大多在廚房發生
10:31
That's where those big peaks are over to the left.
231
616000
3000
即是在左邊那個大尖峰所表示的
10:34
And just for contrast, we can do this with any word.
232
619000
3000
做個比較，也可以爲別的字詞做個風景
10:37
We can take the word "bye"
233
622000
2000
比方說「good bye」裡的
10:39
as in "good bye."
234
624000
2000
這個「bye」
10:41
And we're now zoomed in over the entrance to the house.
235
626000
2000
我們現在放大房子入口部分
10:43
And we look, and we find, as you would expect,
236
628000
3000
我們查找也發現，各位看得出來
10:46
a contrast in the landscape
237
631000
2000
可以作爲對比的風景
10:48
where the word "bye" occurs much more in a structured way.
238
633000
3000
在那兒「bye」的頻率建構出清楚的風景
10:51
So we're using these structures
239
636000
2000
所以我們利用這種風景結構
10:53
to start predicting
240
638000
2000
開始進行預測
10:55
the order of language acquisition,
241
640000
3000
語言習得的先後順序
10:58
and that's ongoing work now.
242
643000
2000
那就是在我們目前的工作
11:00
In my lab, which we're peering into now, at MIT --
243
645000
3000
我在MIT的研究室－即是現在看到的－
11:03
this is at the media lab.
244
648000
2000
那是在媒體實驗室裡
11:05
This has become my favorite way
245
650000
2000
從影片汲取任何空間影像
11:07
of videographing just about any space.
246
652000
2000
已經成爲我最喜歡的方法
11:09
Three of the key people in this project,
247
654000
2000
這個專案有三個主要人物
11:11
Philip DeCamp, Rony Kubat and Brandon Roy are pictured here.
248
656000
3000
即是影像裡的Philip DeCamp、Rony Kubat和Brandon Roy
11:14
Philip has been a close collaborator
249
659000
2000
Philip是大家看到的
11:16
on all the visualizations you're seeing.
250
661000
2000
影片製作的同事
11:18
And Michael Fleischman
251
663000
3000
還有Michael Fleischman是在我實驗室裡的
11:21
was another Ph.D. student in my lab
252
666000
2000
另一位博士生
11:23
who worked with me on this home video analysis,
253
668000
3000
他和我一同分析這支家庭影片
11:26
and he made the following observation:
254
671000
3000
他說了以下的意見：
11:29
that "just the way that we're analyzing
255
674000
2000
他說「我們分析
11:31
how language connects to events
256
676000
3000
語言如何與事件相關聯
11:34
which provide common ground for language,
257
679000
2000
以作爲語言的共通基礎
11:36
that same idea we can take out of your home, Deb,
258
681000
4000
同樣想法也可帶到你家之外，Deb
11:40
and we can apply it to the world of public media."
259
685000
3000
我們可以用它來分析外面世界的公衆媒體」
11:43
And so our effort took an unexpected turn.
260
688000
3000
結果我們的研究有了料想不到的轉折
11:46
Think of mass media
261
691000
2000
想到公衆媒體
11:48
as providing common ground
262
693000
2000
提供共同的基礎
11:50
and you have the recipe
263
695000
2000
那就可以將我們的
11:52
for taking this idea to a whole new place.
264
697000
3000
想法帶到嶄新的境地
11:55
We've started analyzing television content
265
700000
3000
於是我們開始採用相同的原則
11:58
using the same principles --
266
703000
2000
分析電視的內容－
12:00
analyzing event structure of a TV signal --
267
705000
3000
分析電視訊號的事件結構－
12:03
episodes of shows,
268
708000
2000
播出節目的分集、
12:05
commercials,
269
710000
2000
商業廣告、
12:07
all of the components that make up the event structure.
270
712000
3000
構成事件結構的所有元件
12:10
And we're now, with satellite dishes, pulling and analyzing
271
715000
3000
結果我們現在用衛星碟抽出並分析
12:13
a good part of all the TV being watched in the United States.
272
718000
3000
相當一大部分在美國被觀看的電視節目
12:16
And you don't have to now go and instrument living rooms with microphones
273
721000
3000
現在不用再到各個客廳去裝設麥克風
12:19
to get people's conversations,
274
724000
2000
來取得人們的談話
12:21
you just tune into publicly available social media feeds.
275
726000
3000
只要收聽公衆能取得的社交媒體訊息就行了
12:24
So we're pulling in
276
729000
2000
於是我們每個月抽取
12:26
about three billion comments a month,
277
731000
2000
大約三十億則電視評論
12:28
and then the magic happens.
278
733000
2000
然後美妙的事發生了
12:30
You have the event structure,
279
735000
2000
這當中可以找到事件結構
12:32
the common ground that the words are about,
280
737000
2000
那些字詞內容的共同基礎
12:34
coming out of the television feeds;
281
739000
3000
從這些電視訊息裡透露出來
12:37
you've got the conversations
282
742000
2000
我們取得了關於
12:39
that are about those topics;
283
744000
2000
那些話題的談話
12:41
and through semantic analysis --
284
746000
3000
再經過語意分析－大家現在看到的
12:44
and this is actually real data you're looking at
285
749000
2000
確實是來自於我們進行
12:46
from our data processing --
286
751000
2000
資料處理的真實資料－
12:48
each yellow line is showing a link being made
287
753000
3000
每一條黃線顯示一則評論
12:51
between a comment in the wild
288
756000
3000
在外間造成的連結
12:54
and a piece of event structure coming out of the television signal.
289
759000
3000
於是從電視訊號逐漸顯出一點事件的結構
12:57
And the same idea now
290
762000
2000
同樣的想法現在
12:59
can be built up.
291
764000
2000
可以用來建構關聯
13:01
And we get this wordscape,
292
766000
2000
於是我們得到了這個「字詞風景」
13:03
except now words are not assembled in my living room.
293
768000
3000
只不過這些字詞並非在我家客廳裡組造出來的
13:06
Instead, the context, the common ground activities,
294
771000
4000
而是情境，即共同基礎的活動
13:10
are the content on television that's driving the conversations.
295
775000
3000
即電視的內容在推動談話
13:13
And what we're seeing here, these skyscrapers now,
296
778000
3000
我們現在看到的這些高聳的結構
13:16
are commentary
297
781000
2000
都是電視評論
13:18
that are linked to content on television.
298
783000
2000
在電視內容上有相互關聯
13:20
Same concept,
299
785000
2000
同樣的構想
13:22
but looking at communication dynamics
300
787000
2000
但請看它在另一個
13:24
in a very different sphere.
301
789000
2000
非常不同的空間所造成的溝通動態
13:26
And so fundamentally, rather than, for example,
302
791000
2000
而且深入根本，舉例來說，
13:28
measuring content based on how many people are watching,
303
793000
3000
與測量收視率所得的結果極爲不同
13:31
this gives us the basic data
304
796000
2000
此研究讓我們得到了
13:33
for looking at engagement properties of content.
305
798000
3000
用來檢視內容「佔用特性」的基本資料
13:36
And just like we can look at feedback cycles
306
801000
3000
如同我們可以檢視一個
13:39
and dynamics in a family,
307
804000
3000
家庭裡的反應循環和動態
13:42
we can now open up the same concepts
308
807000
3000
我們現在可以利用同樣的構想
13:45
and look at much larger groups of people.
309
810000
3000
檢視更大的人群
13:48
This is a subset of data from our database --
310
813000
3000
這是從我們資料庫來的一個子集－
13:51
just 50,000 out of several million --
311
816000
3000
只是透過公衆媒體來源取得的
13:54
and the social graph that connects them
312
819000
2000
幾百萬則訊息中的五萬則
13:56
through publicly available sources.
313
821000
3000
以及其間互相關聯的「社交圖」
13:59
And if you put them on one plain,
314
824000
2000
如果把它們放到一個平面上
14:01
a second plain is where the content lives.
315
826000
3000
另一個平面是內容活躍的地方
14:04
So we have the programs
316
829000
3000
於是我們有了節目
14:07
and the sporting events
317
832000
2000
和體育運動事件
14:09
and the commercials,
318
834000
2000
以及商業廣告
14:11
and all of the link structures that tie them together
319
836000
2000
還有所有把它們綁在一起的連結結構
14:13
make a content graph.
320
838000
2000
形成了一個「內容圖」
14:15
And then the important third dimension.
321
840000
4000
然後是重要的第三個面向
14:19
Each of the links that you're seeing rendered here
322
844000
2000
大家在這裡看到的每個連結
14:21
is an actual connection made
323
846000
2000
是某個人說了某東西
14:23
between something someone said
324
848000
3000
和某一件內容之間的
14:26
and a piece of content.
325
851000
2000
真實關聯
14:28
And there are, again, now tens of millions of these links
326
853000
3000
這裡又有這些關聯的幾千萬條連結
14:31
that give us the connective tissue of social graphs
327
856000
3000
這讓我們看見「社交圖」的「關聯組織」
14:34
and how they relate to content.
328
859000
3000
以及它們是如何與內容相關的情況
14:37
And we can now start to probe the structure
329
862000
2000
我們現在可以開始用
14:39
in interesting ways.
330
864000
2000
有趣的方式來探索這個結構
14:41
So if we, for example, trace the path
331
866000
3000
例如，如果我們追蹤
14:44
of one piece of content
332
869000
2000
某一件內容的路徑
14:46
that drives someone to comment on it,
333
871000
2000
那個內容讓某個人對它評論
14:48
and then we follow where that comment goes,
334
873000
3000
然後我們隨著那個評論的走向
14:51
and then look at the entire social graph that becomes activated
335
876000
3000
然後檢視整個啓動的「社交圖」
14:54
and then trace back to see the relationship
336
879000
3000
然後又回頭追蹤查看那個「社交圖」
14:57
between that social graph and content,
337
882000
2000
和內容之間的關係
14:59
a very interesting structure becomes visible.
338
884000
2000
於是顯現出一個非常有趣的結構
15:01
We call this a co-viewing clique,
339
886000
2000
我們稱之爲「共看集團」
15:03
a virtual living room if you will.
340
888000
3000
要說是一個虛擬客廳也可以
15:06
And there are fascinating dynamics at play.
341
891000
2000
這裡頭上演著引人注目的戲劇
15:08
It's not one way.
342
893000
2000
這不是單向的
15:10
A piece of content, an event, causes someone to talk.
343
895000
3000
一件內容，一個事件讓某個人說了話
15:13
They talk to other people.
344
898000
2000
這讓其他的人有感
15:15
That drives tune-in behavior back into mass media,
345
900000
3000
那就驅動了大衆傳媒的收視行爲
15:18
and you have these cycles
346
903000
2000
於是出現了這樣的循環
15:20
that drive the overall behavior.
347
905000
2000
驅動了整體的收視行爲
15:22
Another example -- very different --
348
907000
2000
另一個例子－情況很不同
15:24
another actual person in our database --
349
909000
3000
我們的資料庫裡有一位人士－
15:27
and we're finding at least hundreds, if not thousands, of these.
350
912000
3000
其實我們可以找到成千上百個例子
15:30
We've given this person a name.
351
915000
2000
我們給這位人士一個名字
15:32
This is a pro-amateur, or pro-am media critic
352
917000
3000
這是一位專業業餘者，或專業美國媒體評論員
15:35
who has this high fan-out rate.
353
920000
3000
此人有高度的粉絲收視率
15:38
So a lot of people are following this person -- very influential --
354
923000
3000
許多人追隨這位人士－他很有影響力－
15:41
and they have a propensity to talk about what's on TV.
355
926000
2000
那些追隨者傾向於在電視上說話
15:43
So this person is a key link
356
928000
3000
那麽這位人士是關聯大衆傳媒
15:46
in connecting mass media and social media together.
357
931000
3000
和社交媒體的一個主要連結
15:49
One last example from this data:
358
934000
3000
這份資料的最後一個例子是：
15:52
Sometimes it's actually a piece of content that is special.
359
937000
3000
有時確實是一件特別的內容
15:55
So if we go and look at this piece of content,
360
940000
4000
因此我們現在來檢視這一件內容
15:59
President Obama's State of the Union address
361
944000
3000
才幾個星期前的歐巴馬總統
16:02
from just a few weeks ago,
362
947000
2000
國情咨文演講
16:04
and look at what we find in this same data set,
363
949000
3000
再檢視我們在這組資料中發現些什麽
16:07
at the same scale,
364
952000
3000
用同樣的尺度來衡量
16:10
the engagement properties of this piece of content
365
955000
2000
這件內容的「佔用特性」
16:12
are truly remarkable.
366
957000
2000
真是令人驚奇
16:14
A nation exploding in conversation
367
959000
2000
整個國家頓時同步
16:16
in real time
368
961000
2000
爆發了談話
16:18
in response to what's on the broadcast.
369
963000
3000
那是對廣播的訊息所做出的反應
16:21
And of course, through all of these lines
370
966000
2000
當然，所有這些連結線也
16:23
are flowing unstructured language.
371
968000
2000
也流動著缺乏結構的語言
16:25
We can X-ray
372
970000
2000
我們可以在「社交圖」上
16:27
and get a real-time pulse of a nation,
373
972000
2000
透視一下
16:29
real-time sense
374
974000
2000
在不同的圈子裡
16:31
of the social reactions in the different circuits in the social graph
375
976000
3000
這個被這件內容啓動的國家
16:34
being activated by content.
376
979000
3000
有怎樣的即時脈動和即時官感
16:37
So, to summarize, the idea is this:
377
982000
3000
總結來說，我們的想法是這樣的：
16:40
As our world becomes increasingly instrumented
378
985000
3000
正當我們的世界變得越來越工具化
16:43
and we have the capabilities
379
988000
2000
我們有能力搜集
16:45
to collect and connect the dots
380
990000
2000
並在人們說了些什麽
16:47
between what people are saying
381
992000
2000
和他們說話的情境之間
16:49
and the context they're saying it in,
382
994000
2000
將那些點連結起來
16:51
what's emerging is an ability
383
996000
2000
那麽呈現的將會是洞悉
16:53
to see new social structures and dynamics
384
998000
3000
社交結構和社交動態的新視野
16:56
that have previously not been seen.
385
1001000
2000
那將是前所未有的能力
16:58
It's like building a microscope or telescope
386
1003000
2000
這像是製造了麥克風
17:00
and revealing new structures
387
1005000
2000
和望遠鏡而顯現了
17:02
about our own behavior around communication.
388
1007000
3000
我們的溝通行爲的新結構
17:05
And I think the implications here are profound,
389
1010000
3000
我認爲其中隱含深遠的意義
17:08
whether it's for science,
390
1013000
2000
無論是對科學而言
17:10
for commerce, for government,
391
1015000
2000
對商業而言，對政府而言
17:12
or perhaps most of all,
392
1017000
2000
或也許最重要的是
17:14
for us as individuals.
393
1019000
3000
對我們個人而言
17:17
And so just to return to my son,
394
1022000
3000
那麽我們回到我的兒子
17:20
when I was preparing this talk, he was looking over my shoulder,
395
1025000
3000
我在準備這個談話時，他就在我身後看著
17:23
and I showed him the clips I was going to show to you today,
396
1028000
2000
我讓他看今天給大家看的短片
17:25
and I asked him for permission -- granted.
397
1030000
3000
也徵求他的准許－他同意了
17:28
And then I went on to reflect,
398
1033000
2000
然後我開始醒思
17:30
"Isn't it amazing,
399
1035000
3000
「這不是很令人訝異的嗎？
17:33
this entire database, all these recordings,
400
1038000
3000
這整個資料庫，所有這些錄影紀錄
17:36
I'm going to hand off to you and to your sister" --
401
1041000
2000
我把它們交給你和妹妹」
17:38
who arrived two years later --
402
1043000
3000
妹妹晚了兩年出生
17:41
"and you guys are going to be able to go back and re-experience moments
403
1046000
3000
「你們倆將能夠回顧並重溫
17:44
that you could never, with your biological memory,
404
1049000
3000
你們的生物記憶可能
17:47
possibly remember the way you can now?"
405
1052000
2000
不會記得的那些時刻」
17:49
And he was quiet for a moment.
406
1054000
2000
他沈默了半响
17:51
And I thought, "What am I thinking?
407
1056000
2000
我想「我想到哪裡去了
17:53
He's five years old. He's not going to understand this."
408
1058000
2000
他不過才五歲，不會懂的」
17:55
And just as I was having that thought, he looked up at me and said,
409
1060000
3000
我才剛這麽想，他抬頭看著我
17:58
"So that when I grow up,
410
1063000
2000
說：「那麽，我長大了
18:00
I can show this to my kids?"
411
1065000
2000
可以讓我的孩子看這個？」
18:02
And I thought, "Wow, this is powerful stuff."
412
1067000
3000
我想「哇，這說得可真好」
18:05
So I want to leave you
413
1070000
2000
那麽，我要給各位
18:07
with one last memorable moment
414
1072000
2000
留下最後一個
18:09
from our family.
415
1074000
3000
我們家值得回憶的時刻
18:12
This is the first time our son
416
1077000
2000
這是我兒子第一次
18:14
took more than two steps at once --
417
1079000
2000
走出兩步以上的情況－
18:16
captured on film.
418
1081000
2000
拍攝在影片裡
18:18
And I really want you to focus on something
419
1083000
3000
我真的希望讓大家看的時候
18:21
as I take you through.
420
1086000
2000
要注意到其中一點
18:23
It's a cluttered environment; it's natural life.
421
1088000
2000
周遭有點吵，這是自然的生活環境
18:25
My mother's in the kitchen, cooking,
422
1090000
2000
我媽在廚房做飯
18:27
and, of all places, in the hallway,
423
1092000
2000
就在走道上
18:29
I realize he's about to do it, about to take more than two steps.
424
1094000
3000
我感覺到他就要走出兩步以上
18:32
And so you hear me encouraging him,
425
1097000
2000
因此各位聽到我鼓勵他
18:34
realizing what's happening,
426
1099000
2000
感到有事就要發生
18:36
and then the magic happens.
427
1101000
2000
然後美妙的事發生了
18:38
Listen very carefully.
428
1103000
2000
請仔細聽
18:40
About three steps in,
429
1105000
2000
大概在走三步後
18:42
he realizes something magic is happening,
430
1107000
2000
他感到發生了美妙的事
18:44
and the most amazing feedback loop of all kicks in,
431
1109000
3000
這時最令人訝異的反應循環全都作動了
18:47
and he takes a breath in,
432
1112000
2000
他鬆了一口氣
18:49
and he whispers "wow"
433
1114000
2000
輕輕地說了「哇」
18:51
and instinctively I echo back the same.
434
1116000
4000
我直覺反應地也說了同樣的話
18:56
And so let's fly back in time
435
1121000
3000
我們現在重回那個時光
18:59
to that memorable moment.
436
1124000
2000
回到那個難忘的時刻
19:05
(Video) DR: Hey.
437
1130000
2000
(影片) DR：喂
19:07
Come here.
438
1132000
2000
來，過來
19:09
Can you do it?
439
1134000
3000
你辦得到嗎？
19:13
Oh, boy.
440
1138000
2000
啊，孩子
19:15
Can you do it?
441
1140000
3000
你辦得到嗎？
19:18
Baby: Yeah.
442
1143000
2000
嬰孩：可以
19:20
DR: Ma, he's walking.
443
1145000
3000
DR：媽，他走路了
19:24
(Laughter)
444
1149000
2000
(笑聲)
19:26
(Applause)
445
1151000
2000
(掌聲)
19:28
DR: Thank you.
446
1153000
2000
DR：謝謝大家
19:30
(Applause)
447
1155000
15000
(掌聲)
ABOUT THE SPEAKER
Deb Roy - Cognitive scientistDeb Roy studies how children learn language, and designs machines that learn to communicate in human-like ways. On sabbatical from MIT Media Lab, he's working with the AI company Bluefin Labs.
Why you should listen
Deb Roy directs the Cognitive Machines group at the MIT Media Lab, where he studies how children learn language, and designs machines that learn to communicate in human-like ways. To enable this work, he has pioneered new data-driven methods for analyzing and modeling human linguistic and social behavior. He has authored numerous scientific papers on artificial intelligence, cognitive modeling, human-machine interaction, data mining, and information visualization.
Deb Roy was the co-founder and serves as CEO of Bluefin Labs, a venture-backed technology company. Built upon deep machine learning principles developed in his research over the past 15 years, Bluefin has created a technology platform that analyzes social media commentary to measure real-time audience response to TV ads and shows.
Follow Deb Roy on Twitter>
Roy adds some relevant papers:
Deb Roy. (2009). New Horizons in the Study of Child Language Acquisition. Proceedings of Interspeech 2009. Brighton, England. bit.ly/fSP4Qh
Brandon C. Roy, Michael C. Frank and Deb Roy. (2009). Exploring word learning in a high-density longitudinal corpus. Proceedings of the 31st Annual Meeting of the Cognitive Science Society. Amsterdam, Netherlands. bit.ly/e1qxej
Plenty more papers on our research including technology and methodology can be found here, together with other research from my lab at MIT: bit.ly/h3paSQ
The work that I mentioned on relationships between television content and the social graph is being done at Bluefin Labs (www.bluefinlabs.com). Details of this work have not been published. The social structures we are finding (and that I highlighted in my TED talk) are indeed new. The social media communication channels that are leading to their formation did not even exist a few years ago, and Bluefin's technology platform for discovering these kinds of structures is the first of its kind. We'll certainly have more to say about all this as we continue to dig into this fascinating new kind of data, and as new social structures continue to evolve!
Deb Roy | Speaker | TED.com