シャドーイング練習: Computational Linguistics: Crash Course Linguistics #15 - 動画で英語スピーキングを学ぶ
レッスンを作成中...
1
Hi, I'm Taylor and welcome to Crash Course Linguistics.
2
Computers are pretty great, but they can only do stuff that humans tell them to do.
3
Counterintuitively, this means that the more automatic a human skill is, the more difficult it is for us to teach to computers.
4
It's easy for us to teach a computer to calculate millions of digits of pi, or play chess.
5
But get a computer to recognize which image contains a traffic light?
6
Surprisingly difficult.
7
The same thing goes for language.
8
The parts that are difficult for humans, like learning lots of new words, are easy for computers.
9
Humans, like understanding across typos and accents, or knowing if someone's sad or angry or joking, are really, really difficult for machines.
10
Plus, language isn't just one task to teach.
11
It's all the different things we've talked about throughout the series and more.
12
Programming computers to process human language is called natural language processing, or computational linguistics.
13
We rely on NLP for a whole range of tasks.
14
search engines, voice -activated home systems, spam filters, spell checkers, predictive text, and more.
15
Today we'll look at what NLP is and what makes language a difficult challenge for computers.
16
Getting a computer to work with something as complex as language requires a lot of steps.
17
First, we need to give the computer text to work with.
18
We can input it directly, or get the computer to transform speech sounds, handwriting, or other physical text into digital text.
19
We do that with speech -to -text, handwriting recognition, or optical character recognition processes.
20
This step involves figuring out where the break between words and sentences go, such as the difference between a moist towelette versus a moist towelette.
21
"owlet," or whether a small speck is the dot of an "i," a period, or a flock of dirt.
22
Once it has the digital text, we then need the computer to figure out a the meanings of the words and b the relationship between them.
23
It might use context to disambiguate between things like bank and blank, a riverbank and a financial bank, or common nouns and proper nouns.
24
In this step, the machine figures out approximately what is being said.
25
answer a question, translate it into another language, or find directions between two places.
26
Each of these tasks also requires a different system.
27
All of this data gets produced in some abstract form that the computer understands, like a mathematical equation or some lines of code.
28
The last step is to re -encode that data into natural human language, which can involve text generation.
29
Depending on what the user wants, the computer might need to produce the answer as speech, in which case it would use text -to -speech and speech synthesis.
30
That's a lot of steps!
31
The nice thing about splitting up natural language processing into different steps is
32
that we can reuse parts of it for other tasks.
33
For example, if we make one system that's good at text -to -speech for English, it can read aloud answers to questions, translations into English, and directions to go to places.
34
We can also distinguish between what needs to be customized for each human language and what can always stay in computer code.
35
That saves programmers and computers some time.
36
also be useful by themselves.
37
Automatic captioners may just do the speech -to -text part, screen readers may just do text -to -speech, and search or translation may start with text and skip processing speech entirely.
38
A similar set of steps could work for signed languages too, although this technology is very underdeveloped compared to what's been created for a few big spoken languages.
39
They could be something like signed -to -text, parsing signs, processing the results for a computer to work with, and rendering the output back into signs.
40
that interoperated between signed and spoken languages.
41
For example, a computer could take input in English and translate it to ASL, or vice versa.
42
Just like with the thousands of spoken languages, though, each of the hundreds of signed languages would still need to be supported separately.
43
One thing that won't really help is gloves.
44
Let's head to the thought bubble to pop that bubble.
45
You might have seen hyperbolic headlines about sign language translation gloves in the news throughout the years.
46
They claim that these gloves can translate American Sign Language into English speech by recognizing the wearer's hand shapes.
47
Unfortunately, these glove makers have made several fundamental misunderstandings about how sign languages work.
48
One is that the grammar of signed languages isn't expressed just in the shape of the hand.
49
Signed languages also include facial expressions and movements of the hands and arms in relation to the rest of the body.
50
Two is that signed languages use far more signs than the 26 letters of the manual alphabet, which is all the gloves can detect.
51
Plus, signed languages tend to use the manual alphabet to borrow technical words from spoken languages, not for core vocabulary.
52
That's like making a translation system for English that only recognizes the words that come from Greek.
53
people, but gloves can only translate from signs to speech, never from speech to a format accessible for deaf and hard of hearing people.
54
Which is ironic, because the technology to produce written captions of speech already exists.
55
Computational tools involving signed languages could one day exist, using other input sources that can actually access full signs,
56
but they're never going to be any good if deaf people aren't consulted in creating them.
57
And many deaf researchers have already pointed out that gloves are just never going to accomplish that.
58
So let's say we've created a system that's pretty good at each of the steps involved in natural language processing, at least for one or two languages.
59
Does the system understand language the way a human does?
60
To answer that, let's pretend we've trained a rabbit to press buttons A, B, and C in order to get a treat.
61
We could relabel those buttons "I want food," but that wouldn't mean that the rabbit understands English.
62
The rabbit would press the same buttons if they were labeled something entirely unrelated.
63
The same goes for a computer.
64
If we tell a computer a few basic instructions, it can give the appearance of understanding language.
65
But it might fall apart spectacularly when we ask it to do something more complicated.
66
That's part of what makes teaching a computer to do language so tricky.
67
Originally, people taught computers to do language tasks with long lists of more and more specific rules, such as "make a word plural by adding s," Wait,
68
unless the word is child, in which case add ren instead, and so on, for other exceptions.
69
More modern approaches to machine learning involve showing computers a whole bunch of data to train them on statistical patterns,
70
and then testing how well they've figured out these patterns using a different set of data.
71
A lot of recent leaps in natural language processing have come from a kind of statistical machine learning known as neural networks.
72
Neural nets are based on a very simplified model of how neurons work in the brain, allowing which factors are the most relevant in the training data.
73
But because they work out these factors for themselves, it's hard for humans to know exactly what patterns they're picking up on.
74
Early in a neural net's training, it will make really silly, non -human -like errors, like returning a text EEEEEEE,
75
because it's worked out that E is the most common letter in English writing.
76
The machine will keep adjusting itself based on the training data, though, and eventually it starts returning things that look more like words.
77
Well, almost.
78
In any kind of machine learning, training data is really important, and there are two kinds of data we can use.
79
The first is data with two corresponding parts that have been matched by humans, such as text with audio, words with definitions, questions with answers,
80
sentences with translations, or images with captions.
81
Using parallel data like this is known as supervised learning, and it's great, but it can be hard to find enough data that has both parts.
82
After all, some humans have to create all of these pairs.
83
The second kind of data has only one component, like a bunch of text or audio or video in one language.
84
Using this kind of non -parallel data is known as unsupervised learning.
85
It's much easier to find, but it's harder to use to train a computer, since it has to learn only from half of the pair.
86
So researchers often use a mix of both: a smaller amount of parallel data to get things started, and then a larger amount of non -parallel data.
87
semi -supervised learning.
88
But none of this data just magically appears.
89
It gets created or gathered by humans, and humans have all sorts of biases.
90
Computer science researcher Harini Suresh created a framework to evaluate bias in machine learning.
91
We can use this framework to see how bias affects the language tools we've discussed in this episode.
92
First, historical bias is when a bias in the world gets reflected in the output the computer produces.
93
For example, Turkish doesn't make a gender distinction in any of its pronouns, third -person singular, between he /she /it and singular they.
94
So a translation system might pick a gender for pronouns when translating them from Turkish to English, making "he is a doctor" but "she is a nurse" from the same Turkish pronoun.
95
This might reflect an overall tendency in the world, but our computer is still producing a gender bias.
96
Next, representation bias is when some groups aren't as well represented as others in the training data.
97
For instance, while researchers estimate that at least 2 ,000 languages are actively being used on social media,
98
Only a few large languages are well represented in language tech tools.
99
The rest are barely represented or left out, including all signed languages.
100
When the features and labels in the training data don't accurately reflect what we're looking for, that's measurement bias.
101
The text that has been translated into the most languages is the Bible, so it's often used as training data.
102
But the style of language in religious texts can be very different from day -to -day conversation, and can produce strange results in Google Translate.
103
Aggregation bias is when several groups of data with different characteristics are combined, and a single system isn't likely to work well for all of them at once.
104
If we smushed all the varieties of English into training data for an English speech -to -text program, it could end up working better for standardized English than,
105
say, African American English.
106
program's success based on something users won't find useful.
107
Researchers with an "English first" mentality might focus on whether a predictive text program predicts the next word,
108
whereas the program that predicts the next morpheme would work better for languages with longer words and more morphemes.
109
When a system was originally created for reasonable purposes but then gets misused after its release, that's deployment bias.
110
Style analysis tools can be used to determine whether a historic figure wrote an anonymous anonymous whistleblowers.
111
Being aware of these sources of bias is the first step in figuring out how to correct for them.
112
Like the whole field of computational linguistics, addressing these biases is an active area of research.
113
We have a responsibility to use our increased understanding of language
114
through linguistics to deeply consider the effects we have on each other and the world we live in.
115
This ethical consideration is especially important in computational linguistics because we interact with technology so much in our daily lives.
116
Next time we'll talk about a much older kind of language technology, which is so common, we might not even think of it as a technology.
117
writing systems.
118
If you want to help keep all Crash Course free for everybody, forever, you can join our community on Patreon.
この動画で話す練習をする理由は？
この動画では、計算言語学と自然言語処理について説明しています。言語は非常に複雑であり、コンピュータにとっては挑戦的な分野です。これを理解することで、英語を学ぶ上での深い洞察が得られ、十分な語彙力を身につけることができます。特に、英語の文脈や使い方を学ぶことは、IELTS スピーキング対策にも役立ちます。このような内容を口に出して繰り返すことで、英語の発音を良くするための実践的なステップとなります。 shadowspeak（シャドーイング）を活用して、日常会話や専門的なトピックに触れることができます。
文法と表現の文脈
- 意味の曖昧さを解消する：例えば、「bank」と「blank」の対義語の解釈について。これは語彙の豊富さだけでなく、文脈を理解する力も鍵となります。
- 自動的な処理の重要性：コンピュータが情報を効率的に処理するためには、多段階のプロセスが必要です。話す能力を開発する際には、これらの多層的なアプローチを理解することが重要です。
- コミュニケーションの双方向性：サイン言語と話し言葉をつなぐシステムが求められています。これは、他者との効果的なコミュニケーションを理解し、実践する上での重要な要素です。
よくある発音の落とし穴
この動画で取り上げられているいくつかの用語は、英語学習者にとって発音が難しいかもしれません。例えば、「自然言語処理（natural language processing）」や「計算言語学（computational linguistics）」は、特に非ネイティブスピーカーにとってトラブルの元となります。加えて、アクセントの違いや言葉の区切りなど、言語処理の過程で重要になる要素を練習することで、英語シャドーイングの際に声に出して練習する能力が向上します。
これらの課題に挑むことで、英語の発音やリズムをより良く理解し、友人や同僚との会話をスムーズに進めることが可能になります。shadowspeakを実践することで、英語の習得がより深まるでしょう。
シャドーイングとは？英語上達に効果的な理由
シャドーイング（Shadowing）は、もともとプロの通訳者養成プログラムで開発された言語学習法で、多言語習得者として知られるDr. Alexander Arguelles によって広く普及されました。方法はシンプルですが非常に効果的：ネイティブスピーカーの英語を聞きながら、1〜2秒の遅延で声に出してすぐに繰り返す——まるで「影（shadow）」のように話者を追いかけます。文法ドリルや受動的なリスニングと異なり、シャドーイングは脳と口の筋肉が同時にリアルタイムで英語を処理・再現することを強制します。研究により、発音精度、抑揚、リズム、連音、リスニング力、そして会話の流暢さが大幅に向上することが確認されています。IELTSスピーキング対策や自然な英語コミュニケーションを目指す方に特におすすめです。