Grammatical Categories

Jonathan Robie Jonathan.Robie at SoftwareAG-USA.com
Thu Jul 26 11:08:15 EDT 2001


Ken wrote a very interesting message that explains how he is thinking about 
grammatical categories and meaning quite well, and relates this to computer 
languages.

I currently make a living largely by designing computer languages (see 
http://www.w3.org/tr/xquery), and I will soon be giving a presentation on 
the relationship between syntactic query languages and semantic systems 
like RDF (the talk is called "The Syntactic Web", and I will be presenting 
it in Montreal in mid-August), so I have been thinking about some of the 
issues Ken raises.

In a well designed computer language, there is very little ambiguity in the 
syntax. The languages that I design are LL(1), which means that if you read 
from left-to-right, any given token may be ambiguous, but looking at the 
next one will always clear up the ambiguity. The syntax of the language 
does not tell you the semantics - that's done either in the specification, 
in some formal language such as inference notation, or in code that is 
associated with the productions. However, the semantics can be tightly and 
unambiguously associated with the productions of the grammar.

Neither English or Greek is like that. The form of a sentence is often 
ambiguous, to the great joy of those of us who like puns:

-  I see, said the blind man, who picked up his hammer and saw.
-  Those men are flying machines.
-  I knew a man with a wooden leg named Sam. (What was the name of his 
other leg?)
-  "Your peace is beyond me; I rest in your peace. I hold to the love that 
I cannot grasp."

Or consider the following example from yesterday's email:

   Date: Thu, 26 Jul 2001 03:28:05 -0500
   From: "Travelocity Member Services" <feedback at travelocity.com>
   Subject: Fly for Less this Summer and Fall

These sentences are all ambiguous in their form. A computer parser can not 
tell you what they mean. If I designed a language that had this kind of 
ambiguity, people would send me email notifying me of the bug that I would 
then fix.

What Ken is asking for is quite reasonable - he would like Greek, and 
probably also English, to be a "good language", as computer people define 
this. He wants to know the rules that determine the semantics of the language:

>    Now, if you don't like that position, then provide
>a meaningful, mediating view that enables someone who
>has read a fair amount of different kinds of Greek to
>determine not the category, which is illusory I'm
>told, but the function of the genitive modifier.
>That's all I'm asking for.  Since the syntax isn't
>determinative (and I never, ever urged otherwise), how
>can one determine the semantic significance besides
>trusting one's gut?  I am not being facetious.  I've
>seen lots of posts about the evilness/uselessness of
>grammatical categories.  So I "call" as I gather they
>say in poker (never played it so I don't know).

The categories are a way of explaining, after the fact, how a particular 
person chose to classify an instance of, say, the Genitive. They are 
explanatory tools. If I say that a particular use of the genitive is 
subjective or objective, and you know what those words mean, then you know 
how I have decided to interpret that instance, and you can take a look at 
the text and tell me whether you agree. The fact that these categories have 
names make some people think that they define the True Meaning of Greek in 
some deep way, but telling you the category is really no different from 
saying that this particular instance seems to work like the English phrase 
"Man of Steel" rather than the English phrase "Bucket of Water". In either 
case, you can look at the Greek sentence and legitimately disagree with me.

We are dealing with human language here. And for some deep reason that goes 
beyond anything I can put into words, English or Greek are much better than 
Java or XQuery when we need to express a psalm, a prayer, a parable, or a 
vision.

Jonathan




More information about the B-Greek mailing list