Ajit Narayanan: A word game to communicate in any language

翻译人员: Cici W 校对人员: Chen Yimei

I work with children with autism.

我从事关于自闭症儿童的工作。

Specifically, I make technologies

具体而言，我发明技术

to help them communicate.

来帮助他们交流。

Now, many of the problems that children

许多自闭症儿童所面临的问题，

with autism face, they have a common source,

它们有一个共同的根源，

and that source is that they find it difficult

而这个根源就是他们觉得

to understand abstraction, symbolism.

很难理解抽象概念和符号。

And because of this, they have a lot of difficulty with language.

因此，他们在语言上存在很大的困难。

Let me tell you a little bit about why this is.

让我来告诉你们一些会这样的原因

You see that this is a picture of a bowl of soup.

你看这是一碗汤的图片。

All of us can see it. All of us understand this.

我们都可以看见它。我们都能理解这个。

These are two other pictures of soup,

这里是另外两张汤的图片。

but you can see that these are more abstract

但是你可以看到，这些更加抽象。

These are not quite as concrete.

这些并不那么具体。

And when you get to language,

而当你使用语言的时候，

you see that it becomes a word

你看它变成了一个单词

whose look, the way it looks and the way it sounds,

它看起来，听起来

has absolutely nothing to do with what it started with,

都跟刚开始的或代表的一碗汤，

or what it represents, which is the bowl of soup.

没有半点关系。

So it's essentially a completely abstract,

所以它本质上是完全抽象的，

a completely arbitrary representation of something

一个对现实生活中事物

which is in the real world,

很随意的陈述，

and this is something that children with autism

这才是自闭症儿童

have an incredible amount of difficulty with.

有极大困难的地方。

Now that's why most of the people that work with children with autism --

这就是为什么如今大多数从事自闭症儿童工作的人

speech therapists, educators --

——语言治疗师们，教育者们

what they do is, they try to help children with autism

——他们所做的是帮助自闭症儿童交流，

communicate not with words, but with pictures.

不是通过文字，而是通过图片。

So if a child with autism wanted to say,

于是如果一个自闭症儿童想说，

"I want soup," that child would pick

"I want soup"(我要汤)，那个孩子将会拿起

three different pictures, "I," "want," and "soup,"

三张不同的图片，"I"(我)，"want"(要)和"soup"(汤)

and they would put these together,

他们会将这些拼凑在一起，

and then the therapist or the parent would

然后治疗师或者家长就能理解

understand that this is what the kid wants to say.

这就是孩子想要说的。

And this has been incredibly effective;

而且这已经卓有成效了；

for the last 30, 40 years

在过去的30，40年间

people have been doing this.

人们一直在这样做。

In fact, a few years back,

事实上，几年前，

I developed an app for the iPad

我在iPad上开发了一个app

which does exactly this. It's called Avaz,

就是这样的运作的。它叫做Avaz，

and the way it works is that kids select

它是这样运行的：孩子们选择

different pictures.

不同的图片。

These pictures are sequenced together to form sentences,

这些图片排列成句子，

and these sentences are spoken out.

然后这些句子被读出来。

So Avaz is essentially converting pictures,

所以 Avaz 基本上在转译图片，

it's a translator, it converts pictures into speech.

他是个翻译器，它将图片翻译成语音。

Now, this was very effective.

现在，这已经很有效了。

There are thousands of children using this,

有数千个孩子正在使用这个，

you know, all over the world,

你知道的，遍布全世界，

and I started thinking about

于是我开始思考

what it does and what it doesn't do.

它可以做的和不能做的地方。

And I realized something interesting:

我发现了一些有趣的地方：

Avaz helps children with autism learn words.

Avaz 可以帮助自闭症儿童学习单词。

What it doesn't help them do is to learn

它没做到的是帮助他们学习

word patterns.

文字模式。

Let me explain this in a little more detail.

让我更具体地解释下这个。

Take this sentence: "I want soup tonight."

比如这句话："I want soup tonight"(我要汤今晚)。

Now it's not just the words here that convey the meaning.

它不仅仅是文字传达了意思。

It's also the way in which these words are arranged,

这些文字的排列方式也起到了作用，

the way these words are modified and arranged.

文字被修饰和排列的方式。

And that's why a sentence like "I want soup tonight"

那就是为什么一句话比如"I want soup tonight"(我要汤今晚)

is different from a sentence like

不同于这句

"Soup want I tonight," which is completely meaningless.

"Soup want I tonight"(汤要我今晚)，后者是没意义 的。

So there is another hidden abstraction here

所以这里有另一个隐藏的抽象概念

which children with autism find a lot of difficulty coping with,

自闭症儿童很难应对，

and that's the fact that you can modify words

那就是，你可以修辞文字

and you can arrange them to have

你可以对他们进行排序

different meanings, to convey different ideas.

来得到不同的含义，来表达不同的思想。

Now, this is what we call grammar.

这就是我们所说的语法。

And grammar is incredibly powerful,

而且语法极其有用的，

because grammar is this one component of language

因为语法是语言的一个这样的组成部分。

which takes this finite vocabulary that all of us have

它用我们所拥有的有限的词汇

and allows us to convey an infinite amount of information,

使我们表达无限的信息，

an infinite amount of ideas.

无限的思想。

It's the way in which you can put things together

这是一种你可以将事物连在一起

in order to convey anything you want to.

来表达任何你所要表达的。

And so after I developed Avaz,

于是在我开发了 Avaz 之后，

I worried for a very long time

我担心过很长一段时间

about how I could give grammar to children with autism.

关于我该怎样将语法传递给患了自闭症的孩子们。

The solution came to me from a very interesting perspective.

解决方案来自于一个很有意思的观点。

I happened to chance upon a child with autism

我碰巧遇到一个自闭症孩子

conversing with her mom,

正在和他的妈妈交谈，

and this is what happened.

然后发生了这样的事。

Completely out of the blue, very spontaneously,

完全出乎意料的，很自然地，

the child got up and said, "Eat."

那个孩子站起来并说道，"Eat"(吃)

Now what was interesting was

非常有意思的是

the way in which the mom was trying to tease out

这位妈妈试图梳理出

the meaning of what the child wanted to say

孩子所以表达的意思

by talking to her in questions.

通过提问来与她交谈。

So she asked, "Eat what? Do you want to eat ice cream?

于是她问道，“吃什么？你想吃冰淇淋吗？

You want to eat? Somebody else wants to eat?

你想吃？还是别人想吃？

You want to eat cream now? You want to eat ice cream in the evening?"

你现在想吃奶油？还是你晚上想吃冰淇淋？“

And then it struck me that

这件事让我突然想到

what the mother had done was something incredible.

这位妈妈做的事情是不可思议的。

She had been able to get that child to communicate

她已经能够让孩子和她交流想法

an idea to her without grammar.

却不需要语法。

And it struck me that maybe this is what

我突然想到也许这就是

I was looking for.

我一直在寻找的。

Instead of arranging words in an order, in sequence,

与其将文字 按顺序，按词组，按句子排列，

as a sentence, you arrange them

不如将他们安放在

in this map, where they're all linked together

这个地图中，它们都是联系在一起

not by placing them one after the other

不是通过将他们一个接着一个地排放

but in questions, in question-answer pairs.

而是按问题，按 [问题-答案] 成对排列。

And so if you do this, then what you're conveying

如果你这样做，那么你所表达的

is not a sentence in English,

就不是英文中的一个句子，

but what you're conveying is really a meaning,

你所表达的是真实的"意义"，

the meaning of a sentence in English.

英文句子的"意义"。

Now, meaning is really the underbelly, in some sense, of language.

"意义"在某种程度上是语言的软肋。

It's what comes after thought but before language.

它是一种在想法之后语言之前的东西。

And the idea was that this particular representation

所以这个想法是，这种特别的表达方式

might convey meaning in its raw form.

可能能够用原始形式传递"意义"。

So I was very excited by this, you know,

我对此非常兴奋，你知道的，

hopping around all over the place,

手舞足蹈地，

trying to figure out if I can convert

试图确定我是否能够将

all possible sentences that I hear into this.

所有我听到的句子转换成这样。

And I found that this is not enough.

然后我发现这样是不够的。

Why is this not enough?

为什么不够呢？

This is not enough because if you wanted to convey

不够是因为如果你想要表达

something like negation,

某些东西，比如否认，

you want to say, "I don't want soup,"

你会想说，"I don't want soup"(我不想要汤)，

then you can't do that by asking a question.

然而你不能通过提问来做到这个，

You do that by changing the word "want."

你通过改变“want”（要）这个词来做到的。

Again, if you wanted to say,

再者，如果你想要说，

"I wanted soup yesterday,"

"I wanted soup yesterday"(我昨天想要汤)

you do that by converting the word "want" into "wanted."

你是通过将“want”转换成“wanted”做到的

It's a past tense.

它是个过去时态。

So this is a flourish which I added

所以这是一个重要的进度

to make the system complete.

使我完善这系统。

This is a map of words joined together

这是单词彼此联系在一起的一张图

as questions and answers,

通过问题和答案的方式，

and with these filters applied on top of them

通过顶上这些过滤器的应用

in order to modify them to represent

来修改他们来表达

certain nuances.

某种细微差别。

Let me show you this with a different example.

让我通过一个不同的例子来展示给大家。

Let's take this sentence:

就拿这个句子来说吧：

"I told the carpenter I could not pay him."

"I told the carpenter I could not pay him."(我跟木匠说过了我不能付他钱)

It's a fairly complicated sentence.

这是一个相当复杂的句子。

The way that this particular system works,

这个特别的系统工作的方式就是，

you can start with any part of this sentence.

你可以以句子的任意部分开始。

I'm going to start with the word "tell."

我将从"tell"(说)这个词开始。

So this is the word "tell."

这就是“说‘（tell）这个单词。

Now this happened in the past,

这发生在过去，

so I'm going to make that "told."

所以我将把它变成"说过"(told)。

Now, what I'm going to do is,

现在，我将这样做，

I'm going to ask questions.

我将提问，

So, who told? I told.

那么，谁说(told)了？我说(told)了。

I told whom? I told the carpenter.

我跟谁说(told)了？我跟木匠说(told)了。

Now we start with a different part of the sentence.

现在让我们从句子的另一个部分开始。

We start with the word "pay,"

我们从“付钱”（pay）这个词开始，

and we add the ability filter to it to make it "can pay."

我们加入了能力过滤器，将它变成了“can pay”(能够付钱)。

Then we make it "can't pay,"

然后我们将它变成了“can't pay”(不能付钱)

and we can make it "couldn't pay"

而且我们将它变成了“couldn't pay”((过去)不能付钱)

by making it the past tense.

通过将它变成过去时态。

So who couldn't pay? I couldn't pay.

那么谁"couldn't pay"((过去)不能付钱)？我"couldn't pay"((过去)不能付钱)。

Couldn't pay whom? I couldn't pay the carpenter.

不能付钱给谁？我不能付钱付给木匠。

And then you join these two together

然后你讲这两个关联在一起

by asking this question:

通过问这样的一个问题：

What did I tell the carpenter?

我跟木匠说什么了？

I told the carpenter I could not pay him.

我跟木匠说我不能付钱给他。

Now think about this. This is

现在想想这个问题。这是

—(Applause)—

—（掌声）—

this is a representation of this sentence

这是对这个句子的一个描述

without language.

不需要语言。

And there are two or three interesting things about this.

这里有两三个很有意思的现象。

First of all, I could have started anywhere.

首先，我可以从任何地方开始。

I didn't have to start with the word "tell."

我没有必要从"tell"（说）这个单词开始。

I could have started anywhere in the sentence,

我本可以从句子的任何一个地方开始，

and I could have made this entire thing.

也能弄出这整个东西。

The second thing is, if I wasn't an English speaker,

第二点是，如果我不是一个说英语的人，

if I was speaking in some other language,

如果我说某种其他的语言，

this map would actually hold true in any language.

这个地图能够在任何语言中适用。

So long as the questions are standardized,

只要问题能够被标准化，

the map is actually independent of language.

这个地图事实上是独立于语言的。

So I call this FreeSpeech,

我管这个叫 Free Speech（自由说话），

and I was playing with this for many, many months.

我玩这个玩了好多好多个月。

I was trying out so many different combinations of this.

我试过关于这个的好多个不同的组合。

And then I noticed something very interesting about FreeSpeech.

然后我注意到 Free Speech 有一些非常有趣的现象。

I was trying to convert language,

我尝试过将语言进行转换，

convert sentences in English into sentences in FreeSpeech,

将英文句子转换成 Free Speech的句子，

and vice versa, and back and forth.

然后反向转换，然后反复转换。

And I realized that this particular configuration,

我意识到这种特别的结构，

this particular way of representing language,

这种特别的语言表示方式，

it allowed me to actually create very concise rules

它竟然能让我创造非常简明的规则，

that go between FreeSpeech on one side

这种规则能够在 Free Speech 和

and English on the other.

英语之间转换。

So I could actually write this set of rules

因此我事实上能够写下这些

that translates from this particular representation into English.

将特别表示法转换成英语的规则。

And so I developed this thing.

于是我发明了这个东西。

I developed this thing called the FreeSpeech Engine

我发明了这个，叫做 Free Speech 引擎

which takes any FreeSpeech sentence as the input

它将任何 Free Speech 句子作为输入

and gives out perfectly grammatical English text.

然后输出语法完美的英语文本。

And by putting these two pieces together,

通过将这两部分结合在一起，

the representation and the engine,

表示部分和引擎部分，

I was able to create an app, a technology for children with autism,

我得以创造一个 app， 一种专为自闭症儿童而生的技术，

that not only gives them words

它不仅教他们文字

but also gives them grammar.

它也教他们语法。

So I tried this out with kids with autism,

我给患了自闭症的孩子试过这个，

and I found that there was an incredible amount of identification.

我获得了莫大的肯定。

They were able to create sentences in FreeSpeech

他们能够使用 Free Speech 来创造那些

which were much more complicated but much more effective

比起类似英文语句

than equivalent sentences in English,

更复杂却有效多了的语句，

and I started thinking about

然后我开始思考

why that might be the case.

为什么是这样的。

And I had an idea, and I want to talk to you about this idea next.

我有一个想法，我后面想要和大家说说这个想法。

In about 1997, about 15 years back,

大概在1997年，大约15年前，

there were a group of scientists that were trying

有一群科学家他们试图

to understand how the brain processes language,

弄清楚大脑是如何处理语言的，

and they found something very interesting.

然后他们发现了一些有很意思的现象。

They found that when you learn a language

他们发现当你还说小孩时学习一门语言

as a child, as a two-year-old,

比如2岁大，

you learn it with a certain part of your brain,

你使用大脑的特定的部分来学习它，

and when you learn a language as an adult --

当你作为一名成年人来学习它的时候——

for example, if I wanted to learn Japanese right now —

比如，我现在想学习日语——

a completely different part of my brain is used.

我的大脑的一个完全不同的部分被使用了。

Now I don't know why that's the case,

现在我不知道为什么是这样的，

but my guess is that that's because

但是我猜测那是因为

when you learn a language as an adult,

当你作为成年人学习语言的时候，

you almost invariably learn it

你几乎总是在

through your native language, or through your first language.

通过你的母语或者通过你的第一语言在学习它。

So what's interesting about FreeSpeech

所以对于 Free Speech 而言有意思的是

is that when you create a sentence

当你创造一个句子

or when you create language,

或者当你创造语言时，

a child with autism creates language with FreeSpeech,

患了自闭症的孩子通过 Free Speech 创造语言，

they're not using this support language,

他们没有使用这种语言中介，

they're not using this bridge language.

他们没有使用桥梁语言。

They're directly constructing the sentence.

他们直接在构造句子。

And so this gave me this idea.

这给了我这样一个想法。

Is it possible to use FreeSpeech

这样是可能的吗？不仅仅将 Free Speech

not for children with autism

给自闭症儿童使用

but to teach language to people without disabilities?

也给没有缺陷的人学习语言用？

And so I tried a number of experiments.

于是我尝试了一些实验。

The first thing I did was I built a jigsaw puzzle

首先我做的时建造一个拼图游戏

in which these questions and answers

它的问题和答案

are coded in the form of shapes,

以图形和颜色

in the form of colors,

的形式编码，

and you have people putting these together

然后你让人们把他们拼凑在一起

and trying to understand how this works.

然后试图理解这是怎么运作的。

And I built an app out of it, a game out of it,

我通过它制作了一个应用，是一个游戏

in which children can play with words

孩子们可以用文字游戏

and with a reinforcement,

然后巩固，

a sound reinforcement of visual structures,

用声音强化视觉结构，

they're able to learn language.

让他们学习语言。

And this, this has a lot of potential, a lot of promise,

然后这个，这个有很大的潜力，很多承诺，

and the government of India recently

而且印度政府最近

licensed this technology from us,

从我们这项技术授权，

and they're going to try it out with millions of different children

他们将在上百万不同的儿童身上使用

trying to teach them English.

尝试教他们英语。

And the dream, the hope, the vision, really,

而这个梦想，希望，远景，实际上，

is that when they learn English this way,

就是他们能够用这种方式学习英语，

they learn it with the same proficiency

他们想能够将它学得熟练得

as their mother tongue.

跟他们的母语一样。

All right, let's talk about something else.

好吧，让我们来谈点别的东西。

Let's talk about speech.

让我们谈谈语言。

This is speech.

这个是语言，

So speech is the primary mode of communication

语言是我们之间

delivered between all of us.

交流主要模式。

Now what's interesting about speech is that

关于语言有意思的是

speech is one-dimensional.

语言是一维的。

Why is it one-dimensional?

为什么是一维的呢？

It's one-dimensional because it's sound.

它是一维的因为它是声音。

It's also one-dimensional because

它是一维的也是因为

our mouths are built that way.

我们的嘴是那样构造的。

Our mouths are built to create one-dimensional sound.

我们的嘴被构造得生产一维的声音。

But if you think about the brain,

但是如果你想想大脑，

the thoughts that we have in our heads

在我们脑子中的思法

are not one-dimensional.

不是一维的。

I mean, we have these rich,

我的意思是，我们拥有这些丰富的，

complicated, multi-dimensional ideas.

复杂的，多维的思想。

Now, it seems to me that language

现在，对我而言好像是这样的

is really the brain's invention

语言确实是大脑的发明

to convert this rich, multi-dimensional thought

将这种丰富的，多维的想法

on one hand

进行转换，一方面，

into speech on the other hand.

转换成语言，另一方面。

Now what's interesting is that

现在有趣的是

we do a lot of work in information nowadays,

我们如今在做大量的信息化工作，

and almost all of that is done in the language domain.

而且几乎所有的工作都是在语言领域的。

Take Google, for example.

比如Google，打个比方吧。

Google trawls all these countless billions of websites,

Google 网罗了数十亿的网站，

all of which are in English, and when you want to use Google,

都是英文网站，当你想要使用 Google 时，

you go into Google search, and you type in English,

你进入 Google 搜索，然后你输入英文，

and it matches the English with the English.

然后它将英文与英文匹配。

What if we could do this in FreeSpeech instead?

如果我们能够用 Free Speech 来替代这件事会怎样呢？

I have a suspicion that if we did this,

我有一个猜想就是如果我们做到了这个，

we'd find that algorithms like searching,

我们就会发现类似于搜索，

like retrieval, all of these things,

类似于检索，所有这样的算法，

are much simpler and also more effective,

简单得多同时也有效得多了，

because they don't process the data structure of speech.

因为他们不是处理语言的信息结构，

Instead they're processing the data structure of thought.

而是他们处理想法的信息结构。

The data structure of thought.

想法的信息结构。

That's a provocative idea.

这是个令人振奋的主意。

But let's look at this in a little more detail.

但是让我们更仔细的想想这个问题。

So this is the FreeSpeech ecosystem.

这就是 Free Speech 生态体系。

We have the Free Speech representation on one side,

我们把 Free Speech 的表示法放置在一个站点上，

and we have the FreeSpeech Engine, which generates English.

我们有 Free Speech 引擎， 它生成英文。

Now if you think about it,

现在大家思考一下这个，

FreeSpeech, I told you, is completely language-independent.

Free Speech，我已经跟大家讲过，是完全语言独立的。

It doesn't have any specific information in it

他没有任何特定的信息在里面

which is about English.

关于英语的。

So everything that this system knows about English

于是这个系统所知的所有关于英语的

is actually encoded into the engine.

事实上都被编码写入了引擎。

That's a pretty interesting concept in itself.

它本身就是一个非常有趣的概念。

You've encoded an entire human language

你已经将整个人类的的语种编码

into a software program.

写入了一个软件程序中。

But if you look at what's inside the engine,

但是如果你看看引擎里面有什么，

it's actually not very complicated.

他实际上并不是很复杂。

It's not very complicated code.

不是非常复杂的代码。

And what's more interesting is the fact that

更有趣的情况是

the vast majority of the code in that engine

引擎里绝大多数的代码

is not really English-specific.

都不是英文特有的。

And that gives this interesting idea.

这给出的这样一个有趣的想法。

It might be very easy for us to actually

或许对我们来说

create these engines in many, many different languages,

用许多不同的语言来创造这些引擎是很容易的，

in Hindi, in French, in German, in Swahili.

比如用印地语，法语，德语，斯瓦希里语。

And that gives another interesting idea.

这给了我们另一个有趣的想法。

For example, supposing I was a writer,

比如，假设我是一名作家，

say, for a newspaper or for a magazine.

比方说，为报社或者为杂志工作。

I could create content in one language, FreeSpeech,

我可以用一种语言创作内容，Free Speech，

and the person who's consuming that content,

看内容的人，

the person who's reading that particular information

阅览特定信息的人，

could choose any engine,

能够选择任意引擎，

and they could read it in their own mother tongue,

他们能够用他们的母语来阅读，

in their native language.

用他们的当地语言。

I mean, this is an incredibly attractive idea,

我的意思是，这是一个极其诱人的想法，

especially for India.

特别是对印度而言。

We have so many different languages.

我们有这么多不同的语言。

There's a song about India, and there's a description

有一首关于印度的歌，其中有一段描述

of the country as, it says,

是关于这个国家的，它是这样唱的，

(in Sanskrit).

（梵文）

That means "ever-smiling speaker

意思是说“永远微笑着的

of beautiful languages."

美好的语言的述说者”

Language is beautiful.

语言是美好的。

I think it's the most beautiful of human creations.

我认为它是人类发明中最美好的。

I think it's the loveliest thing that our brains have invented.

我认为他是我们的大脑创造的最可爱的东西。

It entertains, it educates, it enlightens,

它使人欢乐，教导众生，启发心灵

but what I like the most about language

但是关于语言我最喜欢的

is that it empowers.

是它带给人力量。

I want to leave you with this.

我想以一下内容作为结束。

This is a photograph of my collaborators,

这是一张我的同事的照片，

my earliest collaborators

我最早的同事

when I started working on language

当我开始研究语言

and autism and various other things.

和自闭症和许多其他的东西的时候。

The girl's name is Pavna,

这个女孩的名字叫 Pavna，

and that's her mother, Kalpana.

这个是她的妈妈， Kalpana。

And Pavna's an entrepreneur,

Pavna 是一名创业者，

but her story is much more remarkable than mine,

但她的故事比我的更值得一提，

because Pavna is about 23.

因为 Pavna 大约只有23岁。

She has quadriplegic cerebral palsy,

她患有脑性四肢瘫痪，

so ever since she was born,

所以从她出生以来，

she could neither move nor talk.

她既不能行动也不能说话。

And everything that she's accomplished so far,

她目前的所有成就，

finishing school, going to college,

完成学业，进入大学，

starting a company,

创立公司，

collaborating with me to develop Avaz,

与我合作开发 Avaz，

all of these things she's done

她所做的所有这些

with nothing more than moving her eyes.

都是通过移动她的眼睛来完成的。

Daniel Webster said this:

Daniel Webster 曾经这样说过：

He said, "If all of my possessions were taken

他说，“如果我所拥有的一切都将被带走

from me with one exception,

只能有一个例外，

I would choose to keep the power of communication,

我将选择留下交流的力量，

for with it, I would regain all the rest."

因为有了它，我将能够重建所有其他的东西。”

And that's why, of all of these incredible applications of FreeSpeech,

这就是为何，在 Free Speech 所有这些不可思议的应用中，

the one that's closest to my heart

最接近我的心灵的那一个

still remains the ability for this

仍旧是它赋予

to empower children with disabilities

自闭症儿童

to be able to communicate,

能够交流的力量，

the power of communication,

交流的力量，

to get back all the rest.

重建所有其他的东西。

Thank you.

谢谢大家。

(Applause)

（掌声）

Thank you. (Applause)

谢谢。（掌声）

Thank you. Thank you. Thank you. (Applause)

谢谢大家。谢谢。谢谢。（掌声）

Thank you. Thank you. Thank you. (Applause)

谢谢大家。谢谢。谢谢。（掌声）