Re: Syntactic Differentiation of Adverbial vs. Adjectival Adpositions

On Sat, Sep 6, 2008 at 9:20 AM, R A Brown <ray@...> wrote:
> Logan Kearsley wrote:
[...]
>> but it's different from having two distinct classes that each do one
>> thing.
>
> Or would a distinction between adpositions that head phrases used
> adverbially & phrases used adjectivally be shown rather by a difference
> of inflexion?
Could be. That's the 'obvious' way to do it, anyway; same basic idea
as just having two different semantic classes for adverbials and
adjectivals.

> In which case are not such adpositions rather to be
> considered sub-classes of adverbs & adjectives respectively?
Only if you also make a distinction between adverbs and adjectives.
Which would make sense, but isn't necessary.
If I were to collapse adpositions into the same classes as
adverbs/adjectives, though, I'd probably analyze it in the opposite
direction, with adverbs/adjectives being considered 0-valent
adpositions. (Or 1-valent adpositions, if you want to consider the
head of an adposition-modified phrase to be a subject argument to the
adposition, in which case normal adpositions would be 2-valent.)

> For example, let us suppose that Esperanto did mark prepositions in this
> way; then we might have:
> *mi mangxis la fruktojn sure la tablo = I ate the fruit [while I was] on the
> table.
>
> *mi mangxis la fruktojn surajn la tablo = I ate the fruit [which was] on the
> table.
>
> (Again I assume that 'fruit' in English is being used as a mass noun. I
> believe - tho i could well be wrong - that 'frukto' is a count noun in
> Esperanto.)
I think I might end up using something like this in one of my langs;
not *exactly* like this, 'cause the language in question doesn't have
adjectives or adverbs or prepositions, using verbs for all of those
functions instead. But that makes it very natural to just use the verb
conjugation system to the same effect- the verb "to be on" would
conjugate differently to take the action nominal as a subject
(adverbial case) or the noun (adjectival case).

>> There are natlangs with mixed adpositional systems, aren't there?
[...]
> Some languages have a greater number of adpositions occupying either
> preposited or postposited positions. By AFAIK this is due to the way the
> words developed from whatever they'd originally been (often adverbs) to
> their status as adpositions, and that in these languages certain words were
> (almost) always prepositions & others postpositions.
So, it's just a matter of the definition of each adposition whether it
happens to go in front or in back; there's no systematic variation in
syntax.

>> I started contemplating altering one of my conlangs to use this sort
>> of system (it would be a great post-fact historical explanation for
>> why a few irregular features are the way they are), but I have a
>> nagging feeling that it could result in different ambiguity as to
>> which noun is supposed the object of an adposition;
>
> That had actually occurred to me also.
I still haven't been able to think of a good example for this, or
completely convince myself that it's not really a problem.
The only real structural ambiguity I can think of would be in a case
like "I ate the fruit on the table" where either 'fruit' is the object
of post-positional 'on' or 'table' is the object of prepositional
'on'. But here, pragmatics tells you that you can't eat a table, which
nicely resolves the situation. I'm having a hard time coming up with
an example where that sort of ambiguity actually matters, and I'm
pretty sure that it's completely solvable by just using an
adpositional case distinct from accusative or dative. (So, if you
really did want to eat a table on some fruit, you would say "I-NOM ate
the fruit-AD on the table-ACC".)

> phil@PHILLIPDRISCOLL.COM wrote:
>> Logan Kearsley writes:
>>>
>>> Additional thought- how common is it to have adpositional phrases
>>> which can behave adjectivally at all?
>
> Quite common, I think. They were there in Ancient Greek, e.g. hoi en te:i
> ne:i andres = the in the boat people men = the men in the boat.
>
> One could also omit the noun and just use the article if the meaning was
> clear, e.g. hoi en te:i ne:i = the [people] in the boat
>
> Although Classical Latin was strict in using prepositional phrases only
> adverbially, in Late Latin & Medieval Latin they were certainly also used
> adjectivally. AFAIK the use of prepositional phrases both adjectivally &
> adverbially is fairly commonplace in (western?) European languages.
Cool. A follow-up question, then- does this commonly produce the kind
of ambiguity that comes up in English, or are there usually other ways
of dealing with it?

>>> English has a habit of eliding
>>> lots of grammatical information like complementizers and relative
>>> pronouns and copulas in relative clauses,
>>> and so it just occurred to me that every instance of a prepositional
>>> phrase modifying a noun, like "the fruit on the table", could be
>>> explained as a relative clause that's been heavily elided- "the fruit
>>> [which is] on the table".
>
> It could - it is not necessary IMO.
Probably not. Time for some historical linguistics research, I think;
'twould be interesting to find out if that actually is the origin of
adjectival adpositions, and if not, where do they really come from.

>> Good question. In English, participles usually come
>> before the noun. "I ate the stolen fruit." But if
>> there is more to the participle, it comes after: "I
>> ate the fruit stolen by my uncle." This could be
>> taken as an ellipsis of "I ate the fruit which was
>> stolen by my uncle."
>
> It could - but that analysis falls down, I think, in languages where
> participles are clearly marked with adjective endings.
Do such languages commonly exhibit the same kind of inversion, though?
Actually, even if they do, that doesn't mean anything; it could just
as well be a difference between "heavy" vs. "light" modifiers. A
participle without arguments would be classed along with light
adjectives, and so come beforehand, whereas a participle with
arguments would be classed with heavy modifiers like relative clauses
and so come afterwards, even if they historically are unrelated to
relative clauses.

>> I think someone mentioned a conlang that has a semantic distinction
>> between adverbial and adjectival prepositions; that would be
>> interesting to investigate.
>
> Check Konya and Ilomi.
That's probably what I was thinking of.

-l.