On Sat, 18 Nov 2006 16:56:31 +0100, Henrik Theiling <[log in to unmask]> 

>Harold Ensle writes:
>> I discovered in the analysis, that most definitely, verbs and nouns must
>> exist and are distinct.
>Could you give an outline of your analysis and reasoning?

Actually, I stated it terribly above. And one might question if I
even understand my own in Ankanian there is NO
morphosyntactic differentiation between verbs and nouns.
Originally, I had divided language into semantic elements and
a limited number of logical operators. And what I am really saying
above is that language requires these semantic elements and these
operators. But in natural languages the verb contains a particular
logical operator along with some semantic content. The logical
operator being "=". And just as algerbra requires "=" before it
means anything, so does language require it. In ankanian I divided
it along these lines where essentially the suffix "-i" is in fact
identical to "=". But in order to make the language more comfortable,
I call this operator combined with some semantic content a "verb"
as in natural languages.

>In abstract considerations about engelangs, my conclusion was usually
>the opposite: I had severe problems defining a criterion to
>distinguish verbal and nominal lexicon entries (I am not talking about
>functions in a sentence) that was not totally arbitrary.

The problem is that in a typical lexicon, the operators are often
implicitly included in the item and it is not always obvious where
the boundary exists between the operator and the real semantic content.

>> It turns out, of course, that both of these things are
>> required universally. That is, a language needs a way to express actions 
>> it also needs a way to express the verbal function in a sentence. 
>> for example, in Loglan where a "function" and an "argument" are 
>Anyway, predicates with arguments (=clauses) can be arguments
>themselves: 'I know that you study.'  And there are sentences that
>have arguments only for purely syntactical purposes: 'It is raining.'
>Further, why should having zero arguments define a different word
>class than having one argument, but having two arguments define the
>same as having one?  If you distinguish by argument count, then why
>boil it down to two categories?  What is special about having zero
>arguments?  And what is special about being an argument when the
>default is a nested structure?

Using Loglan as an example was probably not the best as its whole
approach is pretty poor. In my approach the argument is the semantic
content and the function is the operator. In Loglan, they attach
semantics to the operator to make the function (just as in natural
language) But oblique marked nouns also need an operator to relate
them to the function(verb). Loglan makes all of this implicit based
on the position of the argument and the prelearned requirements of
the function. I find this absurd and even natural languages do better,
since they explicitely relate the arguments through case (or other

It seems that above you are trying to define the morphosyntactical
definition of the verb by number of arguments, which I would agree,
would not make sense. (But it is my mislead, considering that I
probably implied that I thought that Loglan was doing something right.)

BTW my functional representation of your above sentences would be:
I know that you study.     knower=I->{studier=you}
It is raining.   thing=rainer->(loc<-)this:time

>My considerations where about both levels, as are yours: morphosyntax,
>and lexicon, e.g. how to map semantics to lexicon entries.  There is
>more than actions and objects, for which the assignment of verbal
>vs. nominal stem is obvious.
>Looking at English verbs, there are really strange ones like 'to have'
>and 'to be', 

'to be' is interesting because it is the semantically "uncolored"
operator "=". 'to have' is another operator, which is sometimes "uncolored"
and in some contexts has some semantic content.

>there are state verbs 'to lie', 'to stand', there are
>events 'to fall', there is 'to like', 'to fail', 'to understand', 'to
>freeze' (intransitive) etc.etc. -- none are actions.  

While Ankanian has no morphosyntactical distinction, it does have a 
lexical distinction used in word formation. objects vs attributes
and attributes are further divided by transient (called action) and
intransient (called descriptor). This also affects some default
behavior in the grammar (another natural language "comfort" issue).

>Then conversion
>from noun to verb and the other way around works on arbitrary
>semantical content -- it is often a purely morphosyntactical operation
>for shaping some lexicon entry to be usable in a certain situation in
>a sentence.
>Then looking at the English lexicon itself, why is 'freeze' the basic
>lexicon entry in English but not 'frozen' (the state)?  

I think this is pretty arbitrary and I don't think there is any right way.
I found past participles as they are used in many languages to
be a real headache and struggled to incorporate it into a logical language.
I essentially gave up and simple treated it as a distinct semantic
element (albeit derived).

>OTOH, why is
>'hot' (the state) the basic entry but not 'make hot' or 'become hot'
>(mere temperature!)?  And corresponding to 'cold' you have the extry
>lexicon entry 'to chill'.  This is often the case, so I'd unify
>'speed' vs. 'fast' in an engelang, too.
>From all this I found that having one open word class feels easiest
>for my engelangs, and I also do not distinguish 'noun' or 'verb' on
>morphosyntax level, where the distinction is given as it is: to be an
>argument and/or to have an argument/several arguments.  Qyn|gi
>demonstrates this quite clearly: lexicon entries are inflected by
>prototypical verbal *and* nominal categories like case, mood/evidence,
>semantical valence, syntactical valence.  They can be arguments and
>have arguments, and both, and none.

I understand why you do not distinguish them, but I think that as such
that a verb includes the operator, it can be morphosyntactically