Ming Yin Lun

What happens when we talk to AI? If you’ve read our first article, you already know why AI seems like a person. Today, we’ll go one step further and break down the complete mechanism behind this interaction. Have you ever used or played with AI? You type in a question, and it responds instantly. It never interrupts you, never mocks you (unless you specifically request it), and often gives replies that seem to understand your situation while providing constructive suggestions.

Is “subject” in language really something we can take for granted? Imagine an everyday scenario: you walk into a coffee shop and say to the clerk, “Hot latte, less sugar, to-go.” The clerk nods and replies, “Okay, I’ll get that ready for you,” and you lower your head to scroll through your phone while waiting for your number to be called. It seems so ordinary that it doesn’t require any thought, but have you ever wondered why you so naturally believe that “he understands”?

語言中的「主體」，真的是理所當然嗎？ 想像一個日常情景：你走進一家星巴克，對著店員說了一句「熱拿鐵，少糖，帶走」。店員點點頭、回了句「好，馬上幫您處理」，你便低頭滑手機，等著被叫號取單。 看似平凡到無需思考，但有沒有想過，你為什麼會這麼自然地相信「他聽懂了」？ 傳統語用學會說，因為「語境」和「語用能力」讓溝通順利進行。 但這樣的說法跳過了真正的問題： 「你怎麼知道對方是主體？」 語言互動需要「主體」，這是語言學的預設。然而，這個看似理所當然的前提，其實是被長期忽略的幻影。 在繼續之前，我們需要先釐清一個容易被混淆的地方： 「主體」與「主體性」並不相同。 主體是真實存在的，無論你是否和店員講話，他都在那裡、過著他的人生。 但「主體性」是語言互動中被結構性塑造出來的結果，只有當你開口與他互動時，語言才會讓他在語言場中成為一個「能被理解、能被回應」的對象。 語言不創造主體，但語言創造主體性。 而這種「被創造出來的主體性」，正是語言啟動時最常被忽略，卻最關鍵的隱形結構。 根據語言幻影理論，主體性不是預先存在於語言之外的東西，而是透過語用場被「組裝」出來的。 這個組裝需要兩個零件：...