There are 4 messages in this issue.
Topics in this digest:
1.1. Re: Destroying the noun/verb distinction
From: Logan Kearsley
1.2. Re: Destroying the noun/verb distinction
From: Brian Rice
1.3. Re: Destroying the noun/verb distinction
From: And Rosta
2.1. Re: Fith Texts
From: And Rosta
Messages
________________________________________________________________________
1.1. Re: Destroying the noun/verb distinction
Posted by: "Logan Kearsley" [email protected]
Date: Mon Apr 23, 2012 11:54 pm ((PDT))
On 23 April 2012 22:05, Alex Fink <[email protected]> wrote:
> On Mon, 23 Apr 2012 14:05:49 -0600, Logan Kearsley <[email protected]>
> wrote:
>
>>On 21 April 2012 12:49, Alex Fink <[email protected]> wrote:
>>> I think I follow you. But I'm not sure in what context your assertion
>>> "'[open-class words] all denote predicates' is an erroneous assumption" was
>>> meant to be made. Perhaps it holds of Palno; certainly for natlangs you'd
>>> need to say more than that to account for lexical categories, and for
>>> clausal syntax; but in UNLWS and languages of its body type (2d Livagian,
>>> mutatis mutandis, etc.) all (open-class) words do denote predicates and
>>> this isn't a problem.
>>
>>I mean that, e.g., "red" does not assert a truth value, and there is
>>nothing about the meaning of the word that logically implies that it
>>should. "This is red" asserts a truth value, because "is" acts as a
>>logical predicate. However, we can do some morphological operations on
>>that word so that it doesn't act as a predicate anymore, like "this
>>being red". That phrase describes exactly the same proposition, but
>>not as a predicate.
>
> Those are certainly facts about English, and about languages in which they
> have a parallel. I get suspicious to see morphosyntactic properties of the
> metalanguage, which for all I know might be incidental, being used to say
> something deep about cross-linguistic semantics.
>
>>European languages generally mix up functions and predicates,
>
> Is this actually unique to European languages? Not that European languages
> can't do odd things, but in this situation I'd usually take the bet that it's
> rather that European languages are overstudied, or that the framework being
> used is pervaded by assumptions arising from having grown up among them.
I'm sure it's not. But I haven't seen academic literature on this
particular topic that actually talks about anything other than
Indo-European vs. American Northwest, so I wasn't about to bring other
families into it. I did not mean to imply "only" European languages do
that, just that they do in fact do it, in contrast to Salish languages
which don't seem to.
>>"Open-class words all denote predicates" implies, to me, that the
>>underlying meaning of a noun or an adjective includes the idea of
>>asserting a truth value. That is clearly false.
>
> Perhaps, as Jeff points out, one of our problems is that the word "predicate"
> has been ill deployed.
That's kinda my point. It makes sense if you interpret "linguistic
predicate" to mean "logical relation" rather than "logical predicate".
But assuming that it really means "logical predicate" results in
awkwardness.
> That said, I deny that it's *clearly* false. Maybe I haven't drunk enough
> vaus, but I don't think you've shown that at the deepest semantic level, once
> morphosyntax has dissolved off (whatever can be meant by that!), e.g. "a cat
> wants bacon" must be taken as functioning differently to "something wants
> something, and the former something is a cat, and the latter something is
> bacon". There, I've recast a clause with one predicate to have three, all of
> them pulling their weight and asserting a truth value.
>
> That is, if you're trying to assert that you _can't_ leave behind class
> membership as a noun or an adjective or a verb once you pass to the Platonic
> realm of semantics, I don't buy it. I'd say these are only sensible notions
> once a language has morphosyntax that breathes the life of predication into
> some relations and not others.
No, no! Quite the opposite! I'm trying to assert that you *can* leave
behind class membership, but moreover that that includes leaving
behind inherent predication. The claim was made that "from the
standpoint of formal logic, [nouns, adjectives and verbs] all denote
predicates", and thus Lojban represents them that way. But I don't
think that's what formal logic says. It makes sense to say that the
underlying semantics of all open class words are relations, and
relations can be predicated, and you can build a logical system that
does predicate all of its relations and then just uses variables /
pronouns to bind them together implicitly. But it does not make sense
to say that they are all actually inherently predicates, because I can
clearly conceive of nominal concepts divorced from any implication of
truth value.
>>That is exactly the problem that Palno runs into, and I've devised a
>>bunch of kludgy ways to explicitly alter the return value to be
>>something other than an assertion of truth. It could be done much more
>>elegantly, as it is in UNLWS, if I started over from scratch with an
>>awareness of the separability of relation and predication. Halkomelem
>>does it, e.g., by using pronominals to connect the internal
>>non-predicated semantic components of different predicates (which
>>corresponds to what UNLWS does with connecting glyphs) and by having
>>an explicit nominalization mechanism that forms essentially headless
>>or internally-headed relative clauses that are specified to project a
>>particular referential component rather than a predication (which is
>>mainly what I've specified Palno to do, although I don't like the
>>results very much).
>
> I wouldn't mind examples of the Halkomelem. If by using pronominals you mean
> something like Jeff's recent sketch is doing, with different pronominals for
> different links (and not something in the vein of a noun-class system, or a
> case system), I'm surprised that a natlang would do that.
Sadly, I turned that book back into the library already. And I'm not
sure I understand Jeff's sketch well enough to use that for
comparison. But I can kinda sketch out how it basically works with
English lexis.
So you've got this lexical item <dog>. The first content word in a
clause fills predicate position and the zero suffix indicates a third
person singular subject, so all by itself <dog> means "It is a dog".
Now you've got another lexical item, <feed>. In a one-word sentence,
it too is in initial position and thus predicate position and we can
conjugate it like <feed-trans-1-3> for "I feed it". Now if I want to
put in an actual object phrase so my sentence has more than one word
in it, there's no separate word that just means the noun "dog", but I
can turn <dog> into a nominalized headless relative clause <nom-dog>
for "that which is a dog" and get <feed-trans-1-3 nom-dog>- "I feed
the dog", or "I feed it which is a dog".
> Mm, thanks. Seems to me that one would want evaluation environments and
> evaluating things in them to be very lightweight operations in a language of
> this sort, human or computer. (I wonder what Kernel might look like if it
> had been invented this way from scratch, as opposed to deriving from Scheme.)
Indeed. I've been working on a toy project to implement a Lisp-like
language on top of an extremely minimal vau interpreter, with all the
rest of the language's basic syntactic forms implemented as a standard
library on top of vaus. It turns out that namespaces are actually a
subset of evaluation environments, so you can use vau expressions that
return their bound environment to implement namespaces, so I decided
to borrow C++ namespace syntax for evaluation in an environment; e.g.,
"env::expr" is equivalent to "(eval env 'expr)", which makes
namespaces look familiar and makes evaluation very syntactically
light.
For human languages, my idea was to have namespace-words act as
phrase-level particles that could bind at any point in the
phrase-structure tree all the way up to a full clause, and would
govern the interpretation of anything in their c-command domain unless
overridden by another namespace word at a lower level.
-l.
Messages in this topic (32)
________________________________________________________________________
1.2. Re: Destroying the noun/verb distinction
Posted by: "Brian Rice" [email protected]
Date: Tue Apr 24, 2012 12:24 am ((PDT))
On Fri, Apr 20, 2012 at 11:14 AM, Jim Henry <[email protected]> wrote:
> On 4/17/12, Brian Rice <[email protected]> wrote:
> > and verb and achieve both purposes by inflection. Most nouns become verbs
> > that result in the object, like a "constructor" which may describe
> > different levels of process, that the user can choose for highlight.
> >
> > For a chair or apple, say, the gerund might be "fruiting" or "assembling"
> > on a more primitive part (probably left unmentioned to indicate what
> should
> > be considered obvious).
>
> This is interesting. Most verb-prominent engelangs have their
> nounlike verbs mean "to be an X", while some natlangs have their
> simplest noun-to-verb derivational operation typically mean "to use an
> X in the typical way", e.g. English zero-derivations from tool nouns
> like "hammer". I don't remember seeing a language where the simplest
> operation meant "to make an X".
>
I'm glad you think so. My principle interest (to slightly restate
rhetorically) is in pointing out an aspect of "being" that is "becoming" or
"how it became". I'll cite Martin Heidegger's Introduction to Metaphysics
for reference. The "becoming" might differ drastically based on how the
item figures into the current perspective. "That chair", for example, might
just have been brought in here, or someone might have turned on the light
and then you saw it, or even you were in an unfamiliar room in the dark and
bumped into it. Maybe it even struck your kneecap and elicited swearing. I
think all these senses of becoming deserve some kind of differentiation. Of
course, it's still a chair, but it might get a different article for each
of those.
The "to use" aspect certainly occurred to me as well. I'm not sure why I
have not taken it as seriously, but maybe that's my programming language
bias shining through: a noun with one implicit use is not fundamentally as
interesting, where the bias is the false functional vs. object-oriented
dichotomy.
My gjâ-zym-byn is a noun-prominent engelang; its two most basic verb
> derivations from nouns are the stative "to be an X" and the active "to
> use an X", but there are more complex derivations involving additional
> affixes, e.g. "to have no/some/little/lots of X".
>
> The Carib language Hixkaryana has a nifty derivational operation
> meaning "to have a good/big X".
>
> I suggest you also read, if you haven't already, the thread "Nominals
> in Verb-heavy Languages" from February 27 to March 4 of this year.
Thanks, that's a good pointer because I haven't had time to dig through
archives. I'll read these shortly:
http://listserv.brown.edu/archives/cgi-bin/wa?A1=ind1202d&L=conlang#17
http://listserv.brown.edu/archives/cgi-bin/wa?A1=ind1202e&L=conlang#4
http://listserv.brown.edu/archives/cgi-bin/wa?A1=ind1203a&L=conlang#18
--
> Jim Henry
> http://www.pobox.com/~jimhenry/
>
--
-Brian T. Rice
Messages in this topic (32)
________________________________________________________________________
1.3. Re: Destroying the noun/verb distinction
Posted by: "And Rosta" [email protected]
Date: Tue Apr 24, 2012 3:21 am ((PDT))
neo gu, On 24/04/2012 01:46:
> On Mon, 23 Apr 2012 14:05:49 -0600, Logan Kearsley<[email protected]>
> wrote:
>> interpretation or routing around it. Thus, it seems more accurate to
>> me to say that, e.g., in Lojban all open-class words denote relations,
>> and therefore any of them can be used as predicates.
[...]
> I think there's some confusion regarding the traditional conlangish
> use of the word predicate, which possibly should be "relation"
> instead.
It's not a conlangish use, it's thus in general linguistics too, where
"predicate" often means only "thing that has arguments". "Relation" too tends
to be applied exclusively to binary relations, not relations with less than or
more than two arguments. Logan's terminological distinction is motivated by
utility -- the need for exactitude -- rather than conformity to prevailing
norms of usage.
--And.
Messages in this topic (32)
________________________________________________________________________
________________________________________________________________________
2.1. Re: Fith Texts
Posted by: "And Rosta" [email protected]
Date: Tue Apr 24, 2012 2:46 am ((PDT))
MorphemeAddict, On 24/04/2012 03:55:
> On Mon, Apr 23, 2012 at 5:36 PM, And Rosta<[email protected]> wrote:
>> I think the FIFO aspect of Fith is a bit of a red herring in as much as
>> there is no connection between the FIFO aspect and its fundamental
>> difference from human language.
>
> Is "FIFO" a typo for "LIFO"? Stacks and Fith are LIFO.
Yes, sorry; I meant LIFO.
Logan Kearsley, On 24/04/2012 00:52:
> On 23 April 2012 15:36, And Rosta<[email protected]> wrote:
>> To linguists and philosophers who believe that syntax (i.e. the
>> Meaning end of sentence structure)
>
> Is that a typo for "semantics"?
No. A sentence pairs phonological structure and a syntactic structure. The
phonological structure is interpreted phonetically and the syntactic structure
is interpreted semantically, but I'd see those interpretations as
extralinguistic and therefore outside of sentence structure.
>> A syntactic tree contains semantically interpreted relations between
>> semantically interpreted lexical nodes -- the syntactic tree is what makes
>> sentence meaning a coherent whole, rather than just a jumble of word
>> meanings.
>
> I would clarify that "relations that can be semantically interpreted
> and lexical nodes that can sometimes be semantically interpreted".
Stuff that's not semantically interpreted is the exception, since semantic
interpretability is the functional purpose of syntax.
> While related, semantic structure is a different thing from syntactic
> structure, and every syntactic theory seems to have it's own idea of
> how the syntax-semantics interface functions.
Entirely analogously, every phonological theory seems to have its own idea of
how the phonology--phonetics interface functions (tho the phonological
theories' interface views are better thought out than syntactic theories', I
think).
My assumption that semantic structure is in the LoT is mainstream, for
cognitive models of language, I believe.
>> So psychologically implemented language must involve a parser that (when
>> hearing) outputs a syntactic tree -- input is phonological structure, output
>> is syntactic structure.
>
> That does not follow. It must output a semantic interpretation, which
> I happen to conceive of as a generalized directed graph rather than a
> tree, but even that's just my own introspection talking. But the type
> of data structure that holds semantic interpretation may be entirely
> unrelated to the type of data structure that holds abstract syntax.
It's debatable whether the parser itself should have to deal with the
phonetics--phonology and semantics--syntax interfaces, at least not in the same
module that deals with sentential correspondences between phonological and
syntactic structures. Syntactic structures don't have to be trees, but they
happen to be. As for the content and structure of semantic representations, I
think that's an extralinguistic matter contingent on the species or the
cognitive software of the machine using the language.
Garth Wallace, On 24/04/2012 07:24:
> On Mon, Apr 23, 2012 at 2:36 PM, And Rosta<[email protected]> wrote:
>> R A Brown, On 23/04/2012 13:46:
>>
>>> The point is that _parsing_ a postfix expression inorder to
>>> evaluate it is utterly pointless. How this would apply to
>>> language generally and not the narrow domain of mathematical
>>> expressions is IMO more controversial and more of a
>>> challenge to those who want to try it.
>
>> It's a challenge not just in finding a solution to a defined problem but
>> rather in finding a coherent characterization of the problem in the first
>> place. A processor that can do the semantic equivalent of mathematical
>> evaluation looks to me like it's not linguistic. To linguists and
>> philosophers who believe that syntax (i.e. the Meaning end of sentence
>> structure) is written in the Language of Thought, then I guess the idea
>> would make sense, with the parser evaluating in the LoT as it goes along.
>> But I myself don't believe that language has access to LoT, and of course
>> even if one did, one has scant idea of what it's like, and so can't base a
>> conlang on it.
>
> If you think of the product of evaluation as a visualized scene,
> there's no "Language of Thought" necessary.
No, this would mean that LoT and scene visualization are the same thing, or
that LoT involves scene visualization.
> No trees, or other hierarchical relationships, needed either.
Exactly. That's a good way for humans to try to understand how Fith works (--
in the way that I now understand it to work after discussion with Ray) -- the
output of every step of the parsing operation is a visualized scene, and indeed
each combinatory operation of the parser combines fragments of visualized
scenes.
R A Brown, On 24/04/2012 07:08:
> But Logan's email now makes me wonder if Jeffrey's aim was
> not so much what I have detailed above, but rather to
> produce a speakable version of FORTH!. If this is so, then
> And is right after all that the 'stack conjunctions' are an
> essential part of the language and not incidental.
I had only said that it was only the stack conjunctions that made Fith
weird/counternatural, not that they were an essential part of the language. Now
that my understanding of Fith has changed (thanks to you), they seem even more
incidental. Even if Jeffrey's primary aim was to make something FORTH-like, I
think the thing that makes Fith unlike (human) language is that the parser
operates on semantically-evaluated items.
--And.
Messages in this topic (94)
------------------------------------------------------------------------
Yahoo! Groups Links
<*> To visit your group on the web, go to:
http://groups.yahoo.com/group/conlang/
<*> Your email settings:
Digest Email | Traditional
<*> To change settings online go to:
http://groups.yahoo.com/group/conlang/join
(Yahoo! ID required)
<*> To change settings via email:
[email protected]
[email protected]
<*> To unsubscribe from this group, send an email to:
[email protected]
<*> Your use of Yahoo! Groups is subject to:
http://docs.yahoo.com/info/terms/
------------------------------------------------------------------------
