ฝึกพูดภาษาอังกฤษด้วยเทคนิค Shadowing จากวิดีโอ: it’s boring, but it’s what bursts the ai bubble
เลือกโหมดฝึก
แต่ละโหมดจะเปิดห้องฝึกแบบเต็มจอ
252 ประโยค
หากประโยคสั้นหรือยาวเกินไป กดที่ Edit เพื่อปรับแก้
1
The single biggest threat to America isn't the national debt, it's not the Strait of Hormuz, and it's not even the AI bubble itself.
2
It's energy.
3
But not energy in the way you're probably thinking.
4
It's not blackouts or your monthly utility bill.
5
It's that America has pushed all its chips to the center of the table and bet the entire economy on one wager, that AI arrives on schedule.
6
And energy is the thing that decides the schedule.
7
Because right now, AI feels invincible.
8
The valuations keep growing, the spending isn't slowing down, and the stock market keeps pumping.
9
But if something's going to pop the AI bubble, it's energy.
10
And by the end of this video, I can prove it to you.
11
But first, the size of the wager.
12
Because betting the economy isn't an exaggeration.
13
In 2024, the biggest tech companies spent a little over $200 billion building AI infrastructure.
14
In 2025, that number hit almost $400 billion.
15
This year, their own estimates say more than $700 billion.
16
So the spending isn't just growing, it's nearly doubling every year, and it's that spending that's carrying the economy.
17
The level of concentration we're seeing in these tech stocks isn't subtle, it's practically screaming, it's a bubble.
18
Last quarter AI accounted for more than half of America's real GDP growth.
19
George Saravellos, head of FX research at Deutsche Bank, wrote in a message to clients, AI machines appear to be saving the US economy right now.
20
In the absence of tech-related spending, the US would be close to, or in, a recession this year.
21
But you probably already know all this.
22
You've likely heard AI is a bubble a million times by now.
23
And here's the thing about all those bubble takes.
24
Every single one of them is arguing about the wrong question.
25
They're arguing about whether AI is here to stay, whether it's truly going to revolutionize every industry, and whether a chatbot that can't count the letters in strawberry is maybe about to take your job.
26
But that was never the right question.
27
The real bet isn't whether AI works, it's actually about whether we can even physically deliver it on time.
28
Whether the AI future can arrive fast enough to justify the insane prices we're paying for it today.
29
Because right now, money is disappearing into AI faster than Thanos can snap his fingers.
30
It's the fastest and largest build-out in history, yet nobody's actually making any money on it.
31
Which is fine, for now.
32
Investors will happily fund losses as long as they think a payoff is on its way.
33
But eventually, the investors, the companies, and the lenders, funding all of this, need that payoff to actually show up.
34
Eventually the capital has to come home with a return.
35
That's how the AI bubble pops.
36
It's not when it fails, but when it's late.
37
When the capital starts to get sick of waiting around and decides to go somewhere else.
38
So what decides whether AI shows up on time?
39
And what determines whether these trillions of dollars from big tech will ever see a return on investment?
40
Well, before I get into that, let's take a quick pause.
41
As you've probably heard this a thousand times by now, learn AI or you'll get left behind, but nobody mentions the hard part.
42
ChatGPT is good at one thing, Claude at another, Gemini something else.
43
So you end up subscribed to all of them.
44
At roughly $20 a month each, that's $60 a month total, or $720 a year, all just to try and keep up.
45
Which brings us to the sponsor of today's video, ChatLLM by Abacus AI.
46
They've solved that problem.
47
Chat LLM by Abacus AI puts every top model in one place.
48
GPT 5.6, Claude Fable 5, Gemini 3.1, even Grok and DeepSeek, and when a new model drops,
49
it's instantly available, which means no waiting around and no new subscriptions.
50
You can pick which model to use yourself, or let the Route LLM feature read your prompt and automatically pick the best model for you.
51
And Chat LLM by Abacus AI goes past just chat.
52
It can generate images, video, and even build full apps and websites, with hosting included, no code needed.
53
The price for all this is just $10 a month, less than any single subscription it replaces.
54
If you're interested, you can check out Chat LLM by Abacus AI using the link below, or in the pinned comment.
55
Thank you to Abacus AI for sponsoring this video.
56
Now back to the single wager our entire economy's riding on.
57
This entire bet is trapped inside a race between two clocks.
58
Clock one decides how fast the AI future can actually arrive, how quickly we can build it, Plug it in and turn it on.
59
Clock 2 decides how long we can afford to wait for it.
60
We'll get to that later because it's the scarier of the two.
61
So, let's start with the first one.
62
Whenever a technological breakthrough like AI shows up, Wall Street loves to hand out the same Mark Twain quote as advice.
63
Don't bet on the miners.
64
Invest in the picks and shovels.
65
The phrase is more overused than the man in the arena
66
speech in a one-bedroom San Francisco apartment with four founders vibe coding an AI app.
67
What Wall Street's saying is, don't try to pick the winning chatbot.
68
Just buy the stuff underneath the boom, the data centers, the power, the infrastructure.
69
But here's what nobody mentions when they hand you that advice.
70
America's running out of picks and shovels, because you can't just build a data center.
71
A data center doesn't get electricity by plugging into the grid and asking nicely.
72
It has to join something called an interconnection queue.
73
That's the waitlist to get access to the power grid.
74
Think of it as the DMV, but for electricity.
75
Except instead of measuring the weight in bad attitudes and hours, the interconnection queue measures it in regulatory headaches and years.
76
In America's largest power grid, that weight has stretched from under two years in 2008 to over 8 years today.
77
And the waitlist is only where the problems start, because AI data centers don't use electricity the way an office tower does.
78
Busy during the day, quiet at night, asleep on weekends.
79
Data centers run full blast around the clock every day of the year.
80
They already eat more than 4% of America's electricity.
81
And by 2035, this number is expected to reach a fifth of the country's electricity consumption.
82
And the main workhorse feeding this entire build-out is this alien-looking device, the gas turbine, which is where the problems continue.
83
It's the machine behind roughly 40% of America's electricity, and they're among the most complex pieces of machinery on Earth.
84
Only a handful of companies can actually build one.
85
Three manufacturers control over three-quarters of the global market.
86
One of them is GE Vernova.
87
Its order backlog has ballooned from roughly 80 gigawatts to 116, and its own CEO says the company will be largely sold out through the end of 2028.
88
And in 2024 alone, developers ordered nearly three times more turbine capacity than every factory on Earth could physically produce in a single year.
89
Which means it doesn't matter how many zeros Big Tech puts on the check, the manufacturers are structurally constrained.
90
But it's not just the gas turbines either.
91
Transformers, which are the machines that step power down from the grid, now take over two and a half years to get with an estimated 30% supply shortage.
92
And for new transmission lines, they take 6.5 years on average to complete due to permitting delays, backlogs, and a mountain of other BS.
93
So it's not just a single waitlist that's backed up.
94
It's the entire American energy supply chain.
95
We're seeing tech move at Netflix speeds, and the infrastructure underneath it moving at the speed of blockbuster.
96
Which is why the companies trapped between these two speeds are starting to act desperate.
97
Two developers reportedly paid $25 million, for a turbine, just to reserve a place in line for one.
98
Others are rebuilding retired jet engines to makeshift the power they need.
99
We're seeing grown, profitable companies bolting old aircraft engines to the ground because they're desperate for power.
100
So the obvious fix is the one you're already thinking.
101
Just build more.
102
Big tech is great at throwing money at problems.
103
Why not throw money at the source until the clock speeds up?
104
Well, we already ran that exact experiment.
105
And money lost.
106
Remember Remember that 8 year wait just to plug into the grid?
107
It belongs to a power grid called PJM Interconnection.
108
PJM is the largest grid operator in America, serving 67 million people across 13 states from Chicago to Washington DC, roughly 1 in 5 Americans.
109
If you live anywhere in that footprint, your lights are on because of decisions they make,
110
and PJM's biggest decision of the year happens at an event nobody watches, the capacity auction.
111
Here's how it works.
112
PJM doesn't just buy electricity, it buys guarantees.
113
Once a year, PJM holds an auction to pay power plants to promise they'll be ready with electricity on the worst days of the year.
114
It's essentially a billion dollar pinky promise that there's going to be electricity when there's a January freeze, or an August heat wave.
115
You can think of it like insurance.
116
They pay a premium so the grid doesn't fail when it's needed most.
117
For most of this century, that insurance was dirt cheap, because American electricity demand flatlined for 20 years.
118
It made the capacity auction the single most boring event in American finance.
119
In the auction covering 2024, a promise of power cleared at about $29 per megawatt day.
120
Then, the data centers arrived.
121
Not one at a time, by the dozen.
122
Buildings
123
that each swallow as much electricity as a small city walking into a system designed for a world where demand barely grew.
124
And the data centers don't want power eventually.
125
They want it now, guaranteed, forever.
126
The next auction cleared at just under $270, a more than 830% spike.
127
The regional bill went from $2.2 billion to $14.7.
128
The same promise, seven times the price, the biggest leap since the capacity market was invented in 2007.
129
That got attention.
130
Governors across the region leaned on PJM, federal regulators signed off, and the auction got a ceiling, a legal maximum on what the grid was allowed to pay.
131
The very next auction hit that ceiling.
132
So did the one after it, and the one after that.
133
What looked like a spike turned out to be the new floor.
134
Then came December 2025, and something happened that had never happened in PJM's history.
135
The auction came up short.
136
Each auction is built around a target reserve margin.
137
It's the extra safety cushion of power supply required above expected
138
peak demand to guarantee the grid can handle those worst days of the year.
139
It's basically the backups in case of an emergency.
140
For the first time in history, PJM went shopping with more money than ever before, and paid the maximum price allowed under federal law,
141
but there still wasn't enough power supply to go around.
142
And that's the physical reality of the grid.
143
All that money can't summon a gas turbine, a transformer, or a transmission line out of thin air when the world's manufacturers are booked years out in advance.
144
Which means the grid failed to buy what it needed and then paid full price for what it got.
145
And the bill for all this doesn't just get sent to big tech, it gets sent to you.
146
It flows straight into your electric bills.
147
The average household in Maryland saw their electric bill rise by $21 a month, or nearly $250 per year.
148
Which may not sound like a lot at first, until you remember that's just the average, for one household, in one state.
149
When you remember that PJM covers 13 states and 67 million people, it's a lot.
150
PJM pays an independent watchdog called Monitoring Analytics.
151
They ran the numbers on the December auction and found that of the $16.4 billion the auction will cost,
152
roughly $6.5 billion can be traced directly to data centers.
153
Forty cents of every capacity dollar traced to data centers.
154
Across the last three auctions, the number is more than $23 billion.
155
So that's clock one, the power clock.
156
The infrastructure can't keep up with the demand, money can't speed it up, and there's no clear path to fixing it.
157
These companies just have to wait.
158
Which sounds survivable.
159
Annoying and expensive, but survivable.
160
Because late is better than never.
161
Unless the thing doing the waiting has an expiration date.
162
And spoiler, it does.
163
Which brings us to clock two.
164
The chip clock.
165
The one that determines how long we can afford to wait.
166
Because those data centers are really just massive warehouses for a single product,
167
semiconductor chips, specialized processors that train AI and the most expensive piece of the entire bet.
168
Of the more than $700 billion a year, the biggest chunk is spent on the chips and computing hardware.
169
And there are two problems with these chips.
170
The first is, they don't age well.
171
Think of that one grandma of yours who hasn't worn sunscreen since the 70s.
172
A cutting edge AI chip stays cutting edge for somewhere between two to six years, depending on who you ask.
173
Newer, faster chips arrive, and the old ones slide toward worthless.
174
But the infrastructure those chips are living on lasts for decades.
175
So the chips age like a smartphone, and the infrastructure ages like a bridge.
176
That mismatch has a price tag.
177
If a data center scales back, or its chips go obsolete after a few years, the multi-billion dollar grid equipment built to serve it stays on the utilities' books for decades,
178
and a chunk of that cost gets shifted to the utilities' customers, meaning you.
179
And then there's the second problem, the bigger one.
180
Right now, tech companies are reporting record earnings.
181
On the surface, you could maybe even say, what AI bubble?
182
But Michael Burry, the investor from the movie The Big Short, thinks those record earnings are an illusion.
183
He believes big tech's cooking the books, and like most things in finance, he believes it comes down to sneaky accounting.
184
When a company buys a machine, it spreads the cost across the years the machine stays useful, called the useful life.
185
That schedule is called depreciation.
186
Stretch the useful life out, and your annual depreciation drops, making profits look bigger today.
187
Shrink the years, and your annual depreciation expense rises, lowering your reported profits.
188
Which means one boring accounting guess, how long does a chip stay useful, quietly determines billions in big tech earnings.
189
And the problems arise when nobody agrees on the answer.
190
The tech giants started extending the official useful life of their servers from a 3 or four-year standard to six years.
191
Burry says the real number is much lower.
192
If he's right, the industry is understating depreciation by about $176 billion between 2026 and 2028.
193
By Burry's math, this would overstate Oracle's earnings by nearly 27% by 2028, and Meta's by almost 21%.
194
NVIDIA fired back immediately at this.
195
They said that their customers get four to six years out of these chips just fine.
196
What we're watching is a classic kindergarten game of he said, she said.
197
That's also what makes this interesting, because whether you side with Burry or NVIDIA doesn't actually matter, because the tech companies themselves can't agree on the answer.
198
In January 2024, Amazon decided its servers would last six years instead of five.
199
That single decision reduced its depreciation expense by $3.2 billion and added $2.5 billion to its profit for the year.
200
Twelve months later, Amazon changed its mind.
201
They walked the estimate from six years back down to five for subsets of their servers and networking equipment.
202
The reason Amazon gave in the filing was the increased pace of technology development, particularly in artificial intelligence.
203
And while Amazon was walking some things back, Meta went the opposite direction.
204
Meta extended the life of its servers and network assets to five and a half years.
205
That single change reduced its 2025 depreciation costs by $2.9 billion, and added roughly a dollar to Meta's earnings per share.
206
So, same technology, same time period, two accounting departments, looking at the same silicon, and reaching opposite conclusions.
207
And that's the race we have.
208
There are two clocks ticking.
209
One says new power infrastructure can take 5 to 10 years to build.
210
The other says today's AI chips may only be economically cutting edge for 3 to 6 years, depending on who you ask, of course.
211
If those clocks fall out of sync, billions of dollars of infrastructure could arrive for a very different AI market than the one companies are planning for today,
212
and the race is already being timed.
213
Goldman Sachs looked at how often data centers actually open on schedule.
214
Historically, only about 72% of projects due within the next year showed up on time.
215
When you adjust for today's supply chain and labor problems, Goldman expects about 60% of next year's promised capacity to arrive when promised look two years out,
216
and it drops to 50%.
217
Flip a coin, that's the current odds the AI future arrives on schedule.
218
And Bloomberg's energy research arm found something even uglier.
219
They ran two forecasts side by side.
220
The first asks a simple question.
221
Based on how many AI chips are actually being sold, how much data center capacity should America be building?
222
The second asks a harder one.
223
How much can America actually power?
224
They found that by 2033, the gap will be 63 gigawatts apart.
225
For context, every operational data center in America adds up to a little over 40 gigawatts.
226
So, the gap Bloomberg found isn't a small rounding error.
227
To deliver the future these companies have already paid for, America would have to find a way to power a gap half as large as every data center the country's already built.
228
That took decades to build, on a grid that now has an 8-year waitlist.
229
The math doesn't just look bad, the math looks impossible.
230
Which brings us to the age-old question.
231
How did we get here?
232
How does the entire energy supply chain get this backed up?
233
Well, here's the uncomfortable answer.
234
It's not because AI is failing, it's because it isn't.
235
The demand is painfully real.
236
Everyone wants in.
237
Data centers drove roughly half of America's electricity demand growth last year.
238
The line for power is measured in years because everyone showed up at once.
239
If you've seen my previous videos, it's the flood stage of the capital cycle.
240
Success is what broke the grid.
241
And the chips?
242
How does the most expensive hardware on Earth expire so fast?
243
Again, not because AI stalled, because it didn't.
244
Every new generation of chips buries the one before it.
245
The technology is moving so fast that its own products can't hold their value.
246
Success is what kills the chips.
247
So the better AI does, the more demand floods the grid and the longer clock one takes.
248
The faster AI improves, the quicker the hardware goes stale and the shorter clock two gets.
249
Success winds both clocks against each other.
250
Now if you'd rather get your financial news from people yelling about earnings beats, No hard feelings.
251
But if you want the deeper stuff, the stuff buried in government reports and depreciation schedules, hit subscribe.
252
Because it's those boring documents where the real stories actually live.
ดาวน์โหลดแอป
ทุกอย่างที่ต้องใช้เพื่อพูดได้คล่อง
AI ให้คะแนนการออกเสียงให้คะแนนทุกประโยค
ฝึก IPAเชี่ยวชาญทุกเสียง
คำศัพท์สร้างคลังคำของคุณ
เกมคำศัพท์เล่นไปเรียนไป
ช่องเดียวกัน
B23
The Stock Market Is Being Used as a Dumping Ground
Daily ConversationWorking Adults
C12
To Save the Economy... We’re About to Break It
Daily ConversationWorking Adults
C12
The $2.5 Trillion Credit Wave Nobody Sees Coming
Daily ConversationWorking Adults
C12
if war bad... why stocks go up?
Daily ConversationWorking Adults
C13
The World's Safest Market Is Breaking
Daily ConversationWorking Adults
B22
The Recession's Already Here… Open Your Eyes
Business EnglishDaily ConversationUniversity Students
B25
The New American Housing Crisis Has Landed
Business EnglishUniversity StudentsWorking Adults
B23
The Stock Market is BROKEN (just accept it)
Business EnglishUniversity StudentsWorking Adults
B22
Something Feels Off About 2026...
Business EnglishUniversity StudentsWorking Adults
C111
The AI Bubble Survives on $1.65 Trillion in Hidden Debt
Business EnglishDaily ConversationUniversity Students
B21
Bitcoin Is a Brilliant Scam and I Can Prove It (Again)
Business EnglishUniversity StudentsWorking Adults
B23
Private Credit: The Secret Industry Taking Over Wall Street
Business EnglishUniversity StudentsWorking Adults
B21
Every Type of Investor Explained in 6 Minutes
Business EnglishUniversity StudentsWorking Adults
B25
The SpaceX IPO... It's Worse Than You Think
Business EnglishUniversity StudentsWorking Adults
B23
Why The Dumbest Investors Win (it's not why you think)
Business EnglishWorking Adults
B23
The Inflation Problem... It's Worse Than You Think
Business EnglishDaily ConversationUniversity Students
B22
Crypto Is a Brilliant Scam and I Can Prove It
Business EnglishDaily ConversationUniversity Students
B24
The 1973 Stock Market Crash that Nobody Remembers
Business EnglishUniversity StudentsWorking Adults
B24
The (Inevitable) United States Debt Crisis
Business EnglishDaily ConversationUniversity Students
C14
The AI Bubble… We Need to Talk
Business EnglishUniversity StudentsWorking Adults
FOR YOU
วิดีโอแนะนำ
C17
20 ประโยค
Alexandr Wang on what matters in an age of abundant intelligence
Daily ConversationUniversity StudentsWorking Adults
C11
137 ประโยค
The Race To Power AI Is Reshaping America's Electric Grid
Daily ConversationWorking Adults
![The Great A.I. Debate: Should We Fear the Digital Revolution? [2015] | Intelligence Squared](https://img.youtube.com/vi/QWUeu3TLZN0/mqdefault.jpg)
C11
183 ประโยค
The Great A.I. Debate: Should We Fear the Digital Revolution? [2015] | Intelligence Squared
Daily ConversationUniversity StudentsWorking Adults
C11
623 ประโยค
The Only SaaS Moats Left in 2026
Daily ConversationUniversity StudentsWorking Adults
C12
74 ประโยค
How AI is helping brands identify consumer trends faster
Daily ConversationUniversity StudentsWorking Adults
B22
20 ประโยค
Is the AI Boom Just Another Bubble?
Daily ConversationWorking Adults
C11
85 ประโยค
Significant slowdown in AI payoff could tip economy into a recession, says Apollo Global's Slok
Daily ConversationWorking Adults
C11
150 ประโยค
Anthropic just proved AI isn't getting better
Daily ConversationWorking Adults
C11
40 ประโยค
I MADE $15B WITH CHAT GPT
Business EnglishDaily ConversationWorking Adults
C12
154 ประโยค
歐巴馬談 AI：喊放慢的公司「是真的在擔心自己的產品」【AI 大人說】
Daily ConversationWorking Adults
C13
448 ประโยค
If You Don’t Have One of These Jobs by 2030… You’re Screwed
Daily ConversationWorking Adults
คำศัพท์และข้อสังเกตด้านการพูดสำหรับบทเรียนนี้
บทเรียนฝึกพูดระดับ B2 นี้ใช้วิดีโอ “it’s boring, but it’s what bursts the ai bubble” เป็นสื่อ คำที่ถูกพูดซ้ำบ่อยที่สุด: chip, center, grid, billion, clock วิดีโอนี้มี 252 ประโยค และ 3242 คำ สำหรับฝึกพูดตาม ช่วงที่มีเสียงพูดยาว 17:51 ผู้พูดพูดเร็ว ประมาณ 182 คำต่อนาที จึงมีการเชื่อมเสียงและลดเสียงค่อนข้างมาก มีเพียง 80% ของคำที่อยู่ใน 3,000 คำที่ใช้บ่อยที่สุดในภาษาอังกฤษ คำศัพท์จึงค่อนข้างยาก
คำศัพท์สำคัญในวิดีโอนี้
คำที่ยากที่สุด 15 คำในวิดีโอ พร้อมคำอ่านและความหมาย:
| คำศัพท์ | คำอ่าน | ความหมาย |
|---|---|---|
| grid คำนาม | /ɡɹɪd/ | ตาราง |
| bubble คำนาม | /ˈbʌbəl/ | ฟอง |
| turbine คำนาม | /ˈtɜː.baɪn/ | กังหัน |
| gigawatt คำนาม | /ˈɡɪɡəˌwɒt/ | กิกะวัตต์, จิกะวัตต์ |
| bore คำกริยา | /boɹ/ | เจาะ, ไช |
| investor คำนาม | /ɪnˈvɛs.tə(ɹ)/ | ผู้ลงทุน |
| mathematics คำนาม | /mæθ(.ə)ˈmæt.ɪks/ | คณิตศาสตร์ |
| shovel คำนาม | /ˈʃʌv.əl/ | พลั่ว |
| queue คำนาม | /kju/ | คิว |
| plug คำนาม | /plʌɡ/ | ปลั๊ก, เต้าเสียบ |
| ceiling คำนาม | /ˈsilɪŋ/ | เพดาน |
| trace คำนาม | /tɹeɪs/ | การตามรอย |
| invent คำกริยา | /ɪnˈvɛnt/ | คิดค้น |
| yell คำกริยา | /jɛl/ | ตะโกน |
| recession คำนาม | /ɹɪˈsɛʃn̩/ | การถอย |
กริยาวลีที่คุณจะได้ยิน
| คำศัพท์ | ความหมาย |
|---|---|
| show up คำกริยา | โผล่ |
| leave behind คำกริยา | ลืม |
ไวยากรณ์ในวิดีโอนี้
โครงสร้างที่ผู้พูดใช้บ่อยที่สุด พร้อมคำพูดจริงจากวิดีโอ:
| โครงสร้าง | ในวิดีโอ |
|---|---|
| Passive voice be + กริยาช่อง 3 — เน้นสิ่งที่เกิดขึ้น ไม่ใช่ผู้กระทำ | is trapped · is expected · was invented |
| Present perfect have/has + กริยาช่อง 3 — เหตุการณ์ในอดีตที่ยังเกี่ยวข้องกับปัจจุบัน | has pushed · You've likely heard · you've probably heard |
| Relative clauses who / which + อนุประโยค — ข้อมูลเพิ่มเติมเกี่ยวกับคนหรือสิ่งของ | available, which means · turbine, which is · Transformers, which are |
การออกเสียงที่ควรระวัง
ผู้พูดใช้รูปย่อและรูปลดเสียง 49 ครั้ง เช่น isn't, can't, doesn't ให้พูดแบบสั้นตามที่ได้ยิน
- เสียง “th”: mathematics /mæθ(.ə)ˈmæt.ɪks/, underneath /ˌʌndɚˈniθ/, breakthrough /ˈbɹeɪk.θɹuː/
- เสียง “sh” และ “zh”: depreciation /dɪˌpɹiːʃɪˈeɪʃən/, interconnection /ˌɪn.tɚ.kəˈnɛk.ʃən/, shovel /ˈʃʌv.əl/, specialize /ˈspɛʃəˌlaɪz/, recession /ɹɪˈsɛʃn̩/
- คำยาว — ลงเสียงหนักให้ถูกพยางค์: depreciation /dɪˌpɹiːʃɪˈeɪʃən/, interconnection /ˌɪn.tɚ.kəˈnɛk.ʃən/, mathematics /mæθ(.ə)ˈmæt.ɪks/, developer /dɪˈvɛləpɚ/, machinery /məˈʃiːnəɹi/
เสียงที่คนไทยมักออกเสียงยาก:
- เสียงท้ายคำ — ออกเสียงให้ครบ ไม่เปลี่ยนเป็นตัวสะกดแบบไทย: bubble /ˈbʌbəl/, mathematics /mæθ(.ə)ˈmæt.ɪks/, shovel /ˈʃʌv.əl/, underneath /ˌʌndɚˈniθ/, trace /tɹeɪs/
- /v/ — ไม่ใช่ /w/ ให้ฟันบนแตะริมฝีปากล่าง: investor /ɪnˈvɛs.tə(ɹ)/, shovel /ˈʃʌv.əl/, developer /dɪˈvɛləpɚ/, invent /ɪnˈvɛnt/, vibe /vaɪb/
- /z/ — เสียงก้อง ไม่ใช่ /s/: specialize /ˈspɛʃəˌlaɪz/, arise /əˈɹaɪz/, pause /pɔːz/, freeze /ˈfɹiːz/
วิธีฝึกกับวิดีโอนี้
- ฟังวิดีโอให้จบหนึ่งรอบโดยยังไม่ต้องพูด แล้วจดคำที่ยังไม่รู้จัก
- เริ่มที่ความเร็ว 0.75× พูดตามทีละประโยค แล้วกลับไปใช้ความเร็วปกติเมื่อเริ่มคล่อง
- อัดเสียงตัวเองแล้วเทียบกับต้นฉบับ โดยสังเกตคำอย่าง grid, bubble, turbine เป็นพิเศษ
เทคนิค Shadowing คืออะไร?
Shadowing เป็นเทคนิคการเรียนรู้ภาษาที่ได้รับการรับรองทางวิทยาศาสตร์ พัฒนาขึ้นสำหรับการฝึกนักแปลมืออาชีพ วิธีการนี้เรียบง่ายแต่ทรงพลัง: คุณฟังเสียงภาษาอังกฤษจากเจ้าของภาษาและพูดตามทันที — เหมือนเงาที่ตามผู้พูดด้วยช่วงเวลาห่าง 1-2 วินาที การวิจัยแสดงว่าเทคนิคนี้ปรับปรุงความแม่นยำในการออกเสียง ทำนองเสียง จังหวะ การเชื่อมเสียง การฟังเข้าใจ และความคล่องแคล่วในการพูดได้อย่างมีนัยสำคัญ