[YG Conlang Archives] > [engelang group] > messages [Date Index] [Thread Index] >
Mike S., On 30/09/2012 00:43:
On Sat, Sep 29, 2012 at 6:03 PM, And Rosta <and.rosta@hidden.email <mailto:and.rosta@hidden.email>> wrote:
Mike S., On 29/09/2012 20:52:
> On Sat, Sep 29, 2012 at 2:48 PM, And Rosta <and.rosta@hidden.email <mailto:and.rosta%40gmail.com> <mailto:and.rosta@hidden.email <mailto:and.rosta%40gmail.com>>> wrote:
> 1. word-classes: Interjection, Unary-op, Binary-op, Formula (= Simple Formula), Formula-root
>
> Why would Formula-root be among the word-classes? Isn't the
> distinction between Formula-root and Formula somewhat analogous to
> that between noun phrases and nouns in English?
Highly analogous. The 'structural head' of a 'noun phrase' beginning with an article is the article, and the noun itself can be deeply subordinated, e.g. [a [very [easy [book] [to read]] indeed]]. But the 'distributional head' is the noun; the noun triggers number concord on verbs; the full phrase can alternate with a bare noun. Similarly, in the phrase Jorge calls 'formula', the distributional head is the simple-formula.
Yes, a bare noun can alternate with a full noun phrase just as
simple-formula can alternate with a formula. In these cases, what do
you consider to be the structural head of the noun phrase, and what
do you consider to be the formula-root? I take it the bare noun and
simple formula themselves?
Yes.
The reason why formula-root is a word class is that it captures the notion 'structural head of a phrase whose distributional head is a (simple) formula'.
But intuitively the "real" word-classes are just Interjection,
Formula, Unary-op and Binary-op. In order to be a word class,
Formula-root would subsume (or be subsumed by) one or more of these,
wouldn't it?
Word-classes are ways of stating generalizations across words. It's fine for one class to subsume others. Of the four word-classes the basic rules recognized, three are defined by the number of complements and one by structural-headhood. There's no reason to privilege the former definitional criterion over the latter. In fact, Interjection, Unary-op and Binary-op are completely superfluous; they don't allow for the stating of useful generalizations. Once we bring in the binding rules we would introduce some more motivated word-classes that replace them (e.g. Quantifier).
> 2. A formula-root is a formula or a word whose complement is a formula-root
> 3. Interjections and Formula-roots have no complements.
>
> Would it be more accurate to say:
>
> 2. A formula-root is a formula or (the ?combination of ) any word
> with all its one or more complementary formula-roots>
> ?
It wouldn't be more accurate in the phrasing with "the combination of", since I was using a formulation that uses only terminal nodes. The phrasing "A formula-root is a formula or any word with all its one or more complementary formula-roots" seems to me to be equivalent to the wording I gave.
What confuses me is that (2) seems to say that formula-root can be a
word that has a complement; meanwhile (3) flatly says that formula
root has no complements. In order to reconcile that, I would think
that the formula-root of (2) and (3) is a constituent class that's
possibly but not necessarily a word, but (1) contradicts that idea by
naming formula-root as a word class.
Aha. The reason why you are confused is that I carelessly wrote a load of bollocks -- I am sorry! I should have written: 2. A formula-root is a formula or a word whose complement is a formula-root 3. Interjections and *Formulas* have no complements. (I should add that in actuality the binding rules state better if argument-markers are complements of formula.) --And.