In my last blog I touched the question
whether a computer can give meaning to its manipulations in the sense that the
computer itself understands what it does. For example, the Google translating
machine translates the question “How many members does your family have?” into
Chinese this way: 您家有几口人?(observant readers
of my blog will notice that I had translated this English sentence in a
different way in my blog: see the photo there). But does the Google translating
machine understand the English and Chinese
sentences in the sense that it knows their
meanings? I think that most people
will say: No, a Google translating machine does not understand the meaning of
the sentences that appear on its screen, while man does understand what s/he
says. Unlike a computer, man cannot only translate sentences but also
understand the meanings of the sentences s/he translates. And an experienced translator
knows that often a verbal translation is impossible and chooses a translation
with a meaning that is as close as
possible to the original text. Since translation computers cannot capture
meanings, they often makes stupid mistakes. They “just” translate. Take these
Dutch sentences:
- Toen mijn moeder aan de was was, zag ik twee vliegen vliegen. Daar was
ook een bij bij. Ze vlogen onder de deur door, over de weg weg.
They should be translated as:
- When my mother was doing the laundry, I saw two flies passing by. They
were accompanied by a bee. They passed under the door and flew away over the
road.
I once
tried to translate them with a translating machine and I got this:
- When my
mother was doing the laundry, I saw two flies passing by. They were accompanied
by a bee. They passed under the door and flew away over the road.
I am afraid
that you don’t understand a word of the computer translation. Or rather, you
understand the meanings of the separate words but not the meaning of the
sentences. Apparently the translation machine translated the Dutch sentences
word by word and didn’t understand the pun. Without doubt, in future this
problem will be solved more or less, but there’ll still remain a residual
category of “impossible translations”. Or is translating a matter of “Weak AI”,
as Searle calls it? (see my last blog) But then computers must be able to
“understand” puns.
Does a
human translator better? In principle s/he does but not always. The Chinese
poet Li Shangyin (c812-858) wrote a famously obscure poem, which has been
translated into English in many different versions (now I follow, more or less
verbally, Frith 2007, pp. 163-5). Even the translations of the title are
different, so Frith: “The Patterned Lute”, “The Inlaid Harp”, “The Ornamented
Zither”. In order to illustrate the different ways that the end of the poem has
been translated (or how obscure the poem is), Frith gives three translations of
the last sentence:
- Did it
wait, this mood to mature with hindsight? In a trance from the beginning, then
as now.
- And a
moment that ought to have lasted for ever has come and gone before I knew.
- This
feeling might have become a thing to be remembered, Only, at the time you were
already bewildered and lost.
Each
translator seems to give a different interpretation of the last sentence (and of
the whole poem). Which is the right one? I think we’ll never know, for, as
Fritch explains, “[t]he problem is that we have no direct access to this hidden
meaning ... All we have is the text.”
Actually we
have two problems here: Firstly, there is no context or the context is obscure
to us. Take the word “bat”. It can mean either a mammal species with wings or a
specially shaped piece of wood used for hitting the ball in sports like
baseball or table tennis. From the context we immediately know what is meant.
The other problem is the absence of communication between the interpreter and
the speaker/writer/etc. of a sentence. Frith (p. 165): “... I want to
communicate to you. ... But how can you ever know that the idea in your mind is
the same as the idea in my mind? There is no way you can get into my mind and
compare the ideas directly. Communication is impossible.” It’s what
philosophers call “the other mind problem”: How to get access to the thoughts
of the other? What you see here is that the absence of the other needs not only
be physical but it can also be psychological. However, both context and access
are important if we want to be able to understand and to know that we
understand in the right way. Of course, the access to the other is only
necessary insofar as someone else is involved. But the access to the other can
seldom be complete and also the context is often not fully clear or it is
obscure. If so, then complete understanding or fully grasping the meaning is
hardly possible. What remains then is misunderstanding.
Reference
Chris Frith, Making
up the mind. How the brain creates our mental world. Malden, MA, etc.:
Blackwell Publishing, 2007.
No comments:
Post a Comment