Print

Print


On 8 August 2015 at 01:33, Patrik Austin <[log in to unmask]> wrote:
> I'm not convinced of any counter arguments as of yet. I never tried to make an all-noun language, but now that I look at it, I simply don't see anything more to it.
>
> Of course it can be a matter of definition - what is a part of speech? Wikipedia says: "Commonly listed English parts of speech are noun, verb, adjective, adverb, pronoun, preposition, conjunction, interjection, and sometimes numeral, article or determiner."

That is exactly what it is- a matter of definition. The trick is
convincing people that *your* definition is a sensible one.

> However I did also propose ways to speak FL. For example, I have the sentence "Boy nominative action, hill top accusative action, house accusative action, see action." (This is the regular grammar S -> | aS; as I mentioned earlier, 'top' can be used instead of 'on', and 'action' can be used instead of 'topic'.)
>
> Shouldn't this be pretty straight forward: just show me which of these words are not nouns!

Oh, that's not straightforward *at all*. Or, alternately, it is
trivially straightforward: *none* of them are nouns, because this
language does not conform to the traditional parts of speech for
European-like languages, and properly needs completely different
terminology to describe it so as to avoid confusion with things that
would have the same names but actually behave differently! (This is
why I Capitalize the names of parts of speech for WSL- there are
native terms for them, but using native terms all the time, like
Lojban does, means that nobody can remember them or understand what
you're talking about, and Capitalizing keeps them visually marked so
yo know a WSL Noun is not the same thing as an English noun.)

> Suppose you'll agree at this point that it's all nouns, but someone goes on to argue that it's not unambiguous (whether that was required or not) . That's a mistake because now that we have the sentence, _now_ we'll have to think of how to generate the correct parse tree for it, and we already know exactly what it must look like. The correct parse tree is this: [action [nominative boy] [accusative [top hill] house] see]. If this tree is ambiguous, then there is no unambiguous!

The argument for two parts of speech is thus:
For each word, we need to know one bit of *syntactic* information
about it, entirely independently of its lexical semantics;
specifically, whether it begins a constituent, or ends one. If you
think that that is simply obvious and unambiguous, that is only
because you, as the creator and/or theoretical native speaker, only
know it intuitively and lack conscious access to your implicit
syntactic knowledge.

With that bit of syntactic information, the tree is indeed
unambiguous. But you need that bit per word, and the standard,
traditional view would note that, since there are two classes of words
that have different syntactic behaviors, even though there *may* be no
clear semantic boundaries between them, that we call those two classes
different parts of speech. Or, we can just straightforwardly call them
"lexical classes" in the official analysis, and let other people muddy
the waters with talk of "parts of speech".

My "Intro to Structure" professor once said (paraphrasing generously,
because I don't remember verbatim that far back) that, when it comes
to parts of speech, there are only two kinds of languages: those that
have two parts of speech, and those that have more parts of speech.
And among languages with only two parts of speech, they are almost
always divided into "function" words (which tell you syntactic
information) and "content" words.

On 8 August 2015 at 02:18, R A Brown <[log in to unmask]> wrote:
> On 07/08/2015 17:03, Logan Kearsley wrote:
[snip]
>> This last bit about all leaf nodes being nouns not being
>> the same as everything being a noun does bring up a bit
>> of ambiguity that you can address to better define your
>> position, though. There are, after all, natlangs that
>> are reasonably argued to have *no* nouns, but that
>> doesn't mean they don't have something that could
>> reasonably be called a "noun phrase"- just that there are
>> no *individual words* in the class "noun phrase".
>
> Ah, _words_ - that's another term that's not clearly
> defined.  Are the things separated by white space in written
> French separate words? They certainly do not correspond to
> the phonological words of the language.  Recently And seemed
> to using 'word', partly at least, to mean what I would call
> a morpheme.

I suppose I really should have been more careful about that,
especially seeing as how "where do words start and end" is central to
my very-slowly-progressing thesis research!
And it is indeed defined differently on a language-by-language basis.

> But enough of that - I sort of think I know what you're
> getting at when you write "[t]here are, after all, natlangs
> that are reasonably argued to have *no* nouns", but I may be
> wrong.  Could you point me to one or two of these?

I believe Straits Salish is the typical example, as analyzed by Eloise
Jelinek (http://www.jstor.org/stable/416325). Several Salishan
languages seem to work pretty much the same way. David Gil's analysis
of Riau Indonesian which he uses to make the case for the
Isolating-Monocategorial-Associational categorization of languages
does not so much claim that it has no *nouns* as it simply claims that
there is no way to tell the difference between verbs and nouns.
Jelinek makes the same claim, but in the case of the Salish languages
it's pretty obvious that members of the single class in question
really do consistently act like verbs, while Riau Indonesian is much
more "fluid".

-l.