《以廣交會為窗：中國外貿翻譯蝶變》Canton Fair as a Window: How China's Trade Interpreting Changed Shape

[Zheng Ge's Take]

On 18 October, China News Service in Guangzhou ran a piece titled "The Canton Fair as a Window: The Transformation of China's Trade Interpreting". I read it and then sat for a while without doing anything.

I have worked as an interpreter, and I have been in export for twenty-six years. I know exactly where an interpreter stands in a trade deal. Not a supporting part — the ticket in. Without it you are not even allowed to open your mouth.

The article is about how that ticket went from a thin booklet to a pair of forty-gram glasses. Three leaps, and every one of them jumped at the same time as the size of China's export business.

Act One: the Thirteen Factories, Canton English, and a booklet called Ghost Talk

In the Qing single-port era, Guangzhou was the only port open to foreign trade, and the Thirteen Factories were the hub of world commerce.

There was no standard business English. The Guangzhou merchants worked one out themselves: Cantonese sounds for the pronunciation, grammar stripped to the bone, no elegance required, only usability. People called it Canton English.

"What do you want" came out as "What thing wantchee". Someone put a small booklet together called Gui Hua — Ghost Talk: plain characters, direct sentences, open it and use it.

That crude method was the earliest trade communication system in modern China, and it genuinely carried the busiest commercial route of the Qing period.

After all these years in export, the sentence I dread most is: "My English is not good enough, I cannot do export." The Qing merchants had no textbooks, no exams, no pronunciation classes and no AI, and they did it anyway. An ugly tool still closes the deal.

Act Two: after 1957, from assigned interpreters to the sign-holding crowd

In 1957 the Canton Fair was founded. Foreign-facing staff were scarce, so from the very first session the fair assigned professional interpreters as standard.

In the 1980s reform and opening released the energy in export. More and more salespeople taught themselves a language and doubled as the interpreter.

The 1990s were the crowded years. Foreign buyer traffic surged, and outside the halls a crowd appeared holding up signs listing the languages they spoke and how good they were, waiting to be picked. People called them the sign-holding interpreters.

Mr Huang, a graduate of Guangdong University of Foreign Studies, class of 2002, remembers it like this: "A crowd of young university students, black hair, white shirts, staring at the exhibitors pacing back and forth outside the fence, waiting to be chosen. Every one of us holding up a sign, showing our languages and our level…"

I read that twice. The first time I saw the bustle. The second time I saw who held the pricing power. Black hair and white shirts are the people; "waiting to be chosen" is who sets the price. Back then the interpreter stood outside the fence, and the price was set by whoever stood outside the fence.

Act Three: today the interpreter is no longer outside the fence

This year the interpreter moved inside the hall and took a seat at the booth.

A humanoid robot called Xiao Zhi handles more than eighty languages and switches between Chinese and English on the spot in the exhibition hall.

A dual-screen AI interpreter: the staff member speaks Chinese, and the machine puts out the buyer's mother tongue directly. The report quotes it as "no waiting, no passing a device back and forth, you can talk to each other directly".

Then iFlytek's AI translation glasses: the subtitles are projected onto the lens itself, the whole thing weighs forty grams, and it translates 122 languages live. A Brazilian buyer put one on, immediately talked with the staff in Portuguese, and said he wanted to take this kind of wearable back to his own country to sell.

Forty grams. In the old days I travelled with an interpreter. The interpreter needed meals, a room, a plane ticket, and had moods. Now: forty grams, 122 languages, and it never asks for leave.

Act Four: three layers of what I saw

Layer one: the tool keeps changing, the logic of value does not.

From Ghost Talk to AI glasses, every generation of tool solves the same problem: cut the cost of communicating across languages. The craft era used human memory, the industrial era used a labour market, the intelligent era uses algorithms. What changes is the method, never the problem.

Layer two: AI translation does not solve a language problem. It solves a confidence problem.

Plenty of small and mid-sized export salespeople do not have bad English. They are afraid to open their mouths. Opening your mouth is the highest step of all. AI translation flattens that step: say it wrong and the machine covers for you; say it slowly and nobody laughs. More ordinary people can now walk in without hiring an interpreter as a middleman first.

The effect on China's export business is not "a few thousand yuan of wages saved". It is a bigger base of people doing the work. Once the base grows, everything else changes with it.

Layer three: the translation tool is a thermometer for export capability.

From Canton English to 122 languages in real time is a hundred and some years. In those hundred and some years China's export business went from one street of the Thirteen Factories to the largest goods trader in the world. The tool was pushed out by the volume: do business at a certain size and a tool of that quality turns up at your door.

Which cuts both ways. If your living still depends on "I speak English", be careful. Every generation of tool removes one kind of middleman fee, and the next one it removes might be the one you are charging.

In One Line

Every time the translation tool got smaller, China's export business got bigger. From one booklet called Ghost Talk to a forty-gram pair of glasses, the way Chinese people talk to the world has moved from "teach me how you say it" to "I will say it my way, and you will understand me".

Source: China News Service (Guangzhou), "The Canton Fair as a Window: The Transformation of China's Trade Interpreting", 18 October 2026. Only the facts stated publicly in that report are used here (Mr Huang's recollection and the published specifications of Xiao Zhi, the dual-screen interpreter and the iFlytek AI glasses). Everything else is Hany's own observation.