Re: Language?
From: John Roth (newsgroups_at_jhrothjr.com)
Date: 11/13/04
- Next message: John Roth: "Re: Language?"
- Previous message: Ed Conrad: "FOUR little WORDS but a mouthful"
- In reply to: Pauline M Ross: "Re: Language?"
- Next in thread: Pauline M Ross: "Re: Language?"
- Reply: Pauline M Ross: "Re: Language?"
- Messages sorted by: [ date ] [ thread ]
Date: Sat, 13 Nov 2004 08:42:41 -0600
"Pauline M Ross" <pmross@ross-software.co.uk> wrote in message
news:gjp9p0ps4jilr3u5jua61b9bfuoen0m04i@4ax.com...
> On Fri, 12 Nov 2004 07:58:37 -0600, "John Roth"
> <newsgroups@jhrothjr.com> wrote:
>
>>>>[John]I've been looking at what I call the movement pattern.
>>> [Pauline]This doesn't mean anything to me.
>
>>Well, let's start with an example:
>>"The horse bolted from the burning barn."
>>
>>In that sentence, the words "bolted" and "from"
>>call up the movement template. That is a
>>representation that takes something ("the horse")
>>from point A ("the burning barn") to point B
>>(not specified in this sentence) and also
>>specifies how the something is moving
>>("bolted.")
>
> OK, I see where you are coming from here. This is a totally new (to
> me) way of looking at language, but I can see that it might produce
> interesting results.
You might want to look at George Lakoff (UC-Berkley)
and connectionist grammers. After slogging through a paper
on these suckers I came up with some of these ideas. I
wouldn't want you to think that what I'm talking about
is the same thing he is, though - I haven't read enough of
his stuff to know, expecially on the professional level.
>
>>Early infant learning studies show that infants
>>learn these patterns (and a number of others)
>>before they are at all linguistically competent:
>>therefore they are, in some sense, "builtins"
>>that don't depend on language for their
>>existance.
>
> Well, movement and containers are fairly universal concepts in the
> real world, so that isn't too surprising, although how they determine
> that an infant can recognise these things is a mystery to me.
I find it rather mysterious, too. I read reports on
this occasionally, and I've got an "oh, wow!" reaction
to the experimental methodology in most cases, but
I haven't a clue about how to set up such an experiment
myself.
>>
>>The way I look at language, they are simply
>>the way various representation areas in the
>>brain look to the language facilities. If the
>>two examples above look intuitively obvious,
>>then that's because they are. However, they
>>are intuitively obvious because that's the way
>>our brains are wired.
>
> I can agree with you that such things may be hard wired, but I would
> propose that that is because they are part of the real world (like the
> straight lines and corners that our brains are also hard-wired to
> detect). But whether this is relevant to language is another matter.
Well, yes. They're hard wired because they're relevant to the
real world. The connection to language is that you don't have to
bring in a huge amount of linguistic mechanism to deal with something
that is basically a "primitive" in the structure. You just invoke it.
To use an example from computer languages, I don't have
to invoke a huge amount of mechanism to add two integers;
the hardware does that. It's a built-in. All I have to do is
map the request (x = a + b) to the underlying hardware,
and I'm done. I do have to invoke a lot of mechanism
to concatinate two variable length string objects: there is no
underlying mechanism that does that, so I have to look at the
language system itself to find out how it works.
This is the crux of what I was asking for originally:
since the movement pattern is a built-in, I don't
have to deal with a huge amount of other mechanism;
I can get right down to the most elementary part of
the language structure.
>>> Why is a mutation required to create a 'language pathway'? How is a
>>> language pathway in the brain different from any other kind of
>>> pathway? Did our ancestors not already have the basic equipment to map
>>> words onto the representation areas?
>>
>>No, at least if you go far enough back in the terms
>>"ancestors."
>>
>>There are a great many representation areas in the human brain:
>>I've seen the number 50 quoted in credible sources, but I don't
>>remember whether this is just the visual area or all inclusive. Each
>>of these areas is separate: at some point it requires a separate
>>nerve pathway to make the final connections to each of those
>>distinct areas.
>
> At this point I hear alarm bells going off. Distinct areas? Separate
> nerve pathways? There is a lot of research now which shows that
> although the brain has certain preferences as to where to store what
> information, it is a very fluid process.
Not from what I'm hearing, or rather I should say that
the results on brain plasticity are greatly exaggerated
in the telling.
>>Well, here's one place where I differ radically from the conventional
>>wisdom. I see "symbol mapping" or "symbol processing" as being
>>a red herring of the first water. I don't see anything symbolic about
>>the process of language in any way, shape, manner or form.
>>I am, by the way, a professional software developer so I have
>>a fairly well developed sense of what's "symbolic" processing
>>and what isn't. Language processing in the brain is not symbolic
>>unless you want to extend the term "symbolic" to such an extent
>>that it is no longer useful for reasoning about the matter.
>
> Well, what is a symbol? Dictionary.com has it as "Something that
> represents something else by association, resemblance, or convention,
> especially a material object used to represent something invisible".
> You talk about representation areas and I talk about symbols: I don't
> see the two as being a million miles apart.
They are totally different things. To use a computer analogy again:
a representation area is in the underlying hardware, a symbol is in
the program text (not even in the specific programming language).
> Here's an example: suppose I want to tell you about something, let's
> say a fish. I could draw it (which would be iconic, not symbolic), but
> if I use language I have to use a symbol, the word 'fish'. I convey
> the concept of a fish from my brain to yours via a symbol, a word
> which represents the object. It's symbolic because it's arbitrary: I
> could use 'poisson' or 'iasg' or <insert fish-word of choice> or a
> sign-language sign, but the process of language requires symbols. Your
> representation areas and mappings are internal, but my symbols are the
> external means of communication.
And the reason I dislike that usage is that it's simply too much
of a universal. In that usage, what isn't a symbol? If I talk about
"symbolic processing" I can say things that simply aren't grounded
in any kind of reality. The human brain is, after all, Turing complete,
so you have a great deal of difficulty figuring out whether you're
looking at a program, or whether you're finally getting down to the
underlying hardware.
I'm taking the opposite approach. I'm looking at the underlying
hardware, and asking what I can build on it without creating
a lot of mechanism on top: that is, without having to create
"language libraries."
I could take another lick from (I think) Lakoff and talk about
the competitive versus the cooperative paradigm, but in order
to do that I'd have to build several layers on top of the hardware.
And there isn't any way to measure those layers in the same way
we can look at the hardware with fMRI, EEG, MEG and so forth.
> By the way, I'm a professional software developer too, so we are equal
> as far as that goes, but clearly our definitions of symbolic
> processing are very different :-)
>
>>I hope what I've said above makes a bit more sense.
>
> Yes, and thank you for taking the trouble. I'm still not sure I'm
> getting it, but it's certainly clearer than it was.
Hopefully this has clarified things a bit more.
John Roth
>
> --
> Pauline Ross
>
- Next message: John Roth: "Re: Language?"
- Previous message: Ed Conrad: "FOUR little WORDS but a mouthful"
- In reply to: Pauline M Ross: "Re: Language?"
- Next in thread: Pauline M Ross: "Re: Language?"
- Reply: Pauline M Ross: "Re: Language?"
- Messages sorted by: [ date ] [ thread ]
Relevant Pages
|
