Loading ...
Sorry, an error occurred while loading the content.

204152Re: Languages With Unusual Syntax (was: Universal language....)

Expand Messages
  • Gleki Arxokuna
    Jul 27, 2014
      2014-07-27 12:44 GMT+04:00 Pete Bleackley <pete.bleackley@...>:

      > Well, there's monovalency of course - in iljena, nouns and verbs are
      > morphologically fused, and the syntax operates by clause chaining. You
      > could describe it as having a single part of speech - the noun/verb
      > compound.
      where can i find the latest version of iljena grammar and its dictionary
      apart from frathwiki articles and

      > Pete Bleackley
      > The Fantastical Devices of Pete The Mad Scientist -
      > http://fantasticaldevices.blogspot.com
      > -----Original Message-----
      > From: Logan Kearsley <chronosurfer@...>
      > To: CONLANG@...
      > Sent: Sun, 27 Jul 2014 12:01 AM
      > Subject: Languages With Unusual Syntax (was: Universal language....)
      > On 26 July 2014 13:28, James Kane <kanejam@...> wrote:
      > > Hi all
      > >
      > > I completely agree with Logan in that using the terms 'noun' or 'verb'
      > to describe weird-syntax-lang is completely wrong and quite apparently so.
      > It resembles Māori (which may well be similar to Riau Idonesian but I don't
      > know) in that there are really only two categories of words, bases and
      > particles.
      > >
      > > Noun phrases and predicates do surface in the language, and there are a
      > couple of ways to tell (such as the causative prefix or passive suffix for
      > verbs) if a base can be used as a noun, verb or modifier, but the trouble
      > is that plenty of bases can be used as all three.
      > >
      > > Also, many of the particles can be used in more than one type of phrase:
      > >
      > > kei te kai
      > > at.PRESENT the.SG food
      > > (I don't guarantee correctness)
      > >
      > > might be a predicate 'eating' or the prepositional phrase 'at the
      > food', depending on how its used. The whole thing is further complicated by
      > the fact
      > That does sound similar, although not identical, to Riau Indonesian.
      > Keeping in mind that I do not actually know Riau Indonesian myself,
      > and base my opinion entirely off of the examples David Gil uses to
      > make the arguments for his minimal analysis of its syntax and
      > Monocategorial-Associational theory. It's useful to make a distinction
      > between the syntactic categories of "verb" and "noun". It's possible
      > to have phrases that most definitely act as verbs or nouns, which are
      > composed of words of different categories, while not having any
      > individual words that git those categories, and thus could be said to
      > belong to those parts of speech. Some Salish languages, for example
      > appear to have no nouns, but they have nominalized phrases, which
      > fulfill the role of nouns in syntax. It sounds like Maori might be
      > doing something similar, but with more ambiguity. Somewhere in between
      > the syntactically-strict regime of Salish and the freer R.I. model.
      > According to David Gil, Riau Indonesian (or large parts of it anyway;
      > he acknowledges that there are function words that complicate things,
      > but that significant chunks of discourse get along without using them,
      > and goes on to analyze those chunks of discourse) not only lacks the
      > parts of speech, but doesn't even bother with the syntactic categories
      > at all. The semantics of composite clauses are just determined by
      > association with all of the lexical meanings of the components,
      > followed by contextual filtering. Set semantics seems like an
      > absolutely perfect fit here, although he doesn't use that formalism in
      > any of the papers I've read.
      > What's particularly interesting is that there don't seem to be any
      > human languages that have *less* than 2 parts of speech. Even Riau
      > Indonesian, which doesn't seem to actually *need* any function words,
      > still does have a distinction between open-class content words and
      > closed-class function words. It's entirely possible to design a logic
      > notation (e.g., combinator calculus) which only requires a single
      > syntactic category with equivalence to predicate calculus or set
      > semantics, so why do we never see that in natlangs?
      > I have experimented with designing a loglang with only a single part
      > of speech, and discovered that it quickly becomes extremely difficult
      > to keep track of in my head when sentences start to get remotely
      > non-trivial. That could just be because I couldn't figure out quite
      > how to design it right, but it is suggestive. A further data point is
      > that concatenative programming languages (based on combinator
      > calculus) generally include some minimal additional syntax to allow
      > for abstraction, even though it's not strictly logically necessary,
      > because it just makes it enormously easier to work with (and a lot
      > more though has been put into designing good minimal concatenative
      > programming languages by other smart people than has been put into
      > designing monocategorial loglangs by me).
      > So, it seems to me that at least two parts of speech are needed
      > essentially in order to manage complexity through, essentially,
      > parenthesize / clarifying order of compositional operations. Some of
      > that can be handled by intonation, but intonation patterns are much
      > less robust against noisy transmission channels than distinct function
      > words. That requirement of function words to manage complexity is
      > probably a human language cognitive universal. But, in theory, we
      > might find aliens who really do use a strictly monocategorial
      > language, either because they somehow manage to get by all the time
      > under a maximal ceiling of discourse complexity (like human kids in
      > the telegraphic speech stage), or because their brains are way better
      > at handling deep, complicated parse trees (like the Fith).
      > > Weird-syntax-lang seems similar, but boiled down to the point that there
      > is only a single phrase type as well as only two word types, and I also
      > agree that grammatical categories such as subject, object, aren't apparent
      > either. The semantic categories such as agent, patient etc. still are but
      > that is irrelevant because those are semantic and have nothing to do with
      > syntax.
      > >
      > > Syntactic discussion is still possible, but it won't have the categories
      > that we're used to.
      > Well, I'd say weird-syntax-lang probably has at least two phrase
      > types, counting clauses as phrases, and at least three word types with
      > the propositional operators (logical connectives and modals). So, not
      > quite as minimalist as it could've been, but that wasn't really the
      > point ('cause predicate calculus notation is not minimalist to start
      > with).
      > Referencing David Gil again, while most (all? probably all) human
      > languages have the ideas of predication and attribution, these are not
      > fundamental operations- rather, predication in the linguistic sense
      > and attribution are just the results of different combinations of
      > predication in the logic sense and syntactic projection. With only the
      > structures that are currently defined, it's not possible for
      > weird-syntax-lang to exhibit predication or attribution phenomena
      > because it lacks the syntactic rules necessary to be able to recognize
      > them.
      > One could imagine a language, however, which *allowed* for predication
      > and attribution structures, but still did not *require* them to form
      > complex phrases and clauses. Languages that use noun coordination in
      > place of / in addition to adjectives sort of do this to a minimal
      > degree.
      > One of the traditional banes of a conlangers existence, figuring out
      > relative clauses, would just fall out as an immediate natural
      > consequence of a system that allowed for controlling argument roles
      > assignment and syntactic category projection entirely independently.
      > You could produce sentences that look like they have a traditional
      > subject/verb/object structure in that sort of system, but I would
      > argue that it would still be wrong to analyse the language in those
      > terms, since the appearance of subject-like and verb-like things would
      > not actually be fundamental- they're just an epiphenomenon of the
      > application of the real rules.
      > The List has discussed before the idea of coming up with Lots More
      > than the usual syntactic categories that could be added to make a
      > coherent maximalist, rather than minimalist, language. Bolting some of
      > that thought on top of some of these ideas for eliminating S/O/V
      > structure could result in a really weird alien language.
      > Are there perhaps other ways that we could go about eliminating the
      > traditional subject/object/verb categories?
      > -l.
    • Show all 7 messages in this topic