On 15 Jul 2015 10:18, "Patrik Austin" <[log in to unmask]> wrote:
> I've also found that universal grammar has been used to refer to three
different things
> Anyway, I wish to add a fourth related term: fundamental grammar. There's
a particular regular grammar of the type S ‒> aS that generates a language
that could be called a fundamental language. The grammar is fundamental to
all languages because all formal languages can be derived from this
particular grammar by the means of a rule-by-rule expansion (plus
presumably any language can be formalised).
> The important part is that although it would be possible to keep creating
more and more complex grammars, no matter where we've come, when we go back
down the ladder, rule by rule, we'll always end up with the same
fundamental grammar. To me this is basically the same as a universal
grammar because it can be demonstrated to be the basis of English,
Japanese, Riau, Klingon, Lojban, predicate logic and all logically possible
formal languages.

Hoping I won't regret it, I will risk entering into this discussion again.

I don't understand the relevance of so-called 'formal languages' to 'real
languages' (by which I mean anything that actually is a language, including
conlangs); I don't see what rules for generating strings of letters will
tell us about language, or how they could be the basis of English or
predicate logic (where predicate logic is distinct from a predicate logic
notation). [In saying that I don't understand, I mean to acknowledge that
there might be a failure in my understanding, rather than a nonsensicality
in the ideas I don't understand, but I'd also say I have a comparatively
good understanding of how language works (by 2015 standards), so the
failure of understanding is not due merely to general ignorance.]

Furthermore -- and here I suppose I may have to blame myself for bothering
to repeat myself -- I've made the falsifiable but unrefuted claim that
predicate--argument structure is precisely what syntax must minimally
consist of (or 'encode'), so we already have a clear idea of what syntax is
in its purest form. As for what level of complexity the syntax must have,
that depends on whether it must be able to unambiguously accommodate
predicate--argument structures of unlimited complexity. A syntax that can
do that is way beyond what natlangs can do. But if you simply abandon every
requirement for unambiguity, then you could make do with zero syntax, each
sentence consisting of nothing but a single predicate, with texts
consisting of sequences of such sentences, and predicate--argument
structure inferred pragmatically. So the complexity and unambiguity of
predicate--argument structure in natlangs is limited but not minimal. And
so I don't see how it is possible to ask a meaningful research question
about what syntax can minimally be reduced to, beyond the two answers I've
given (to whit, the unambiguous syntax is of known complexity, but no
natlang is unambiguous, and the ambiguous natlang syntax can be reduced to
zero syntax).

You may prefer to ignore this intervention, which would be fair enough
given that this is not an academic forum.