Pre/Post-Positions: Where's the head?

[p-glyphs]

When I was planning Umu, I decided to go for something loosely head-final, and keep a comment-topic structure.

At first, I opted for prepositions...

comment // topic
on // table

'on' commenting of the topic 'table'.

Now I'm playing with the notion of adpositions as verbs, using the pre/post-verbal position of the "argument" to signal definiteness/indefiniteness (the same way I treat the arguments of "real" verbs)

on table
'on the table'

table on
'on a table'

It kinda works because the farther right you go, the more topical/definite things become.

But now, I've got nouns jumping appositions and also PPs jumping around the verb. I'm thinking about using the PPs relative position to the verb to signal, not definiteness (since this is done internally), but static vs. dynamic distinctions for spatial predicates. Thanks to Alex on the listserv for putting into words my nebulous thought process:

If we analogise newness in discourse to newness as
an actual state of affairs, then if "on.table" is topical, it should be an
old state, i.e. he was on the table already, jumping up and down; if
"on.table" is in the comment, it's free to be a new state, i.e. have the
force of "onto".

But now I worry. Is it practical to expect a pre/post-positional 'argument' to behave differently from (thematic) pre/post-verbal arguments? (expect the pre/post-verbal NP or DP to signal definiteness... but not PPs... cause the adposition acts like a verb and so definiteness is signaled within the PP itself)

Are there NatLangs with dancing pre/post-positions? If yes, what's up there?

Also The head of a PP is a preposition, right? So with my directionality, I should have post-positions... coming last in the phrase. and maybe drop the dance.

Don't get me wrong. I really like the dance. I'm just not sure how to treat this movement and am also worried for when these blocks will go on to build larger structures.

Cheers

[roninbodhisattva]

Also The head of a PP is a preposition, right? So with my directionality, I should have post-positions... coming last in the phrase. and maybe drop the dance.

Yes, the head of an adpositional phrase is the adposition. So a prepositional phrase is headed by a preposition.

[cromulant]

But now I worry. Is it practical to expect a pre/post-positional 'argument' to behave differently from (thematic) pre/post-verbal arguments? (expect the pre/post-verbal NP or DP to signal definiteness... but not PPs... cause the adposition acts like a verb and so definiteness is signaled within the PP itself)

This is more of a C&C question.

Since this is L&L: I've never heard of a language that does either of 1) encoding definiteness of NPs in pre/postverbal position of arguments, or 2) encoding the dynamic/stative distinction in pre/postpositions. This doesn't mean it doesn't exist somehwere in the world's 6000 odd languages, but I, personally, would be surprised.

But I think Alex has explained to you how it can work. There is a certain logic to the topical-stative analogy.

Are there NatLangs with dancing pre/post-positions? If yes, what's up there?

They're called "ambipositons."

And yes, English has one:

he slept through the whole night
he slept the whole night through

But this is a pretty marginal case. I'm not aware of any language that uses ambipositions extensively. WALS mentions languages that use both prepositions and postpositions, but it is clear from their description that they mean that in such languages, a given adposition is one or the other.

I would imagine that a problem with ambipositions is the uncertainty as to whether the ambiposition is acting as a preposition or a postposition, particularly if it appears between two noun phrases.

Also The head of a PP is a preposition, right?

Yes.

So with my directionality, I should have post-positions... coming last in the phrase. and maybe drop the dance.

It's a tendency, not an absolute rule, for head-final languages to have postpositions. Exceptions include Latin, Chinese, Persian, and Navajo. (Though the head-finality of these languages varies. Headedness is often a mixed bag.)

However, the reverse (head-initial languages with postpositions) is extremely rare.

[merijn]

Dutch has a few "dancing adpositions".
There is the pair in het huis meaning "in the house" vs het huis in "into the house".

[Mecislau]

It's not quite the same thing, but in Russian, if the object of a preposition is a quantified noun, the the being quantified can jump to either side of the preposition and quantifier:

через сто лет
čerez sto let
within hundred.ACC year.GEN.PL
"within a hundred years"

лет через сто
let čerez sto
year.GEN.PL within hundred.ACC
"within about a hundred years"

However, this is a part of several broader phenomena dealing with inversion of quantifiers and quantifiees to express approximation, as well as topical fronting of quantified elements and quantifier stranding.

[Miekko]

Are there NatLangs with dancing pre/post-positions? If yes, what's up there?

They're called "ambipositons."

And yes, English has one:

he slept through the whole night
he slept the whole night through

But this is a pretty marginal case. I'm not aware of any language that uses ambipositions extensively. WALS mentions languages that use both prepositions and postpositions, but it is clear from their description that they mean that in such languages, a given adposition is one or the other.

I would imagine that a problem with ambipositions is the uncertainty as to whether the ambiposition is acting as a preposition or a postposition, particularly if it appears between two noun phrases.

Germanic languages has had a lot of ambipositions, but in e.g. Swedish they've been turning into rather boring prepositions.

Finnish has a whole bunch of ambipositions though, but they tend to take the genitive when postpositioned, and the partitive when prepositioned.

[hwhatting]

A case of ambipositions in German:
wegen "because of, for the sake of":
Ich lass mich nur wegen der Kinder / der Kinder wegen nicht scheiden. " I don't get divorced only because / for the sake of the children"
Post-poned wegen is more limited stylistically and mostly means "for the sake of"; e.g. at least I would say Ich kann wegen des Lärm(e)s nicht schlafen "I can't sleep because of the noise" but not *des Lärmes wegen.

[Matt]

Are there NatLangs with dancing pre/post-positions? If yes, what's up there?

They're called "ambipositons."

And yes, English has one:

he slept through the whole night
he slept the whole night through

Wait, I though those were particles. Or at least, that's what my professor calls "off" in sentences like "she turned off the lights" / "she turned the lights off".

[p-glyphs]

Here's a thought...

I treat ambipositions like verbs, marking the definite post-verbally and the indefinite pre-verbally. Then somehow relativize this structure, allowing it to function as the pre/post-verbal argument of a verb higher up the register.

Because definiteness is assigned during VP... before it becomes a 'relativized' PP of a higher register VP... it allows the PP's relative position to the verb to mark static/dynamic spatial distinctions rather than definiteness.

I like this.

Although is this actually relativization? I always saw this process a noun modifier. Here I'm kinda forcing a VP to function nominally.

If you catch this, please share your thoughts.

Initial VP

in street
'it's in the street'

street in
'it's in a street'

VP becomes PP argument of another VP
post-verbal = static

fall [in street] // boy
'the boy fall in the street'

fall [street in] // boy
'the boy falls in a street'

pre-verbal = dynamic

[in street] fall // boy
'the boy falls into the street'

[street in] fall // boy
the boy falls into a street'

[Miekko]

Are there NatLangs with dancing pre/post-positions? If yes, what's up there?

They're called "ambipositons."

And yes, English has one:

he slept through the whole night
he slept the whole night through

Wait, I though those were particles. Or at least, that's what my professor calls "off" in sentences like "she turned off the lights" / "she turned the lights off".

Many linguists think there's no reason to consider them anything than prepositions - do you call verbs without objects particlerbs? No, you call them (intransitive) verbs. The idea that prepositions can be intransitive is not that far off.

[finlay]

Are there NatLangs with dancing pre/post-positions? If yes, what's up there?

They're called "ambipositons."

And yes, English has one:

he slept through the whole night
he slept the whole night through

Wait, I though those were particles. Or at least, that's what my professor calls "off" in sentences like "she turned off the lights" / "she turned the lights off".

I always thought "particle" was a bit of a catch-all term for "little word that I'm not sure what to call". But then, I don't think of these as prepositions, anyway, because it's kind of one-half of a phrasal verb.

[Ser]

Since this is L&L: I've never heard of a language that does either of 1) encoding definiteness of NPs in pre/postverbal position of arguments

Chinese kind of does, where anything pre-verbal is always definite, but post-verbal NPs can be either definite or indefinite. If you want to say a sentence with an indefinite subject e.g. "a dog is in the garden", it's necessary to use the existential verb 有 yǒu, effectively making the subject of your sentence in English a direct object using the construction "there's a dog in the garden".

狗 在 花園 裡
gǒu zài huāyuán li
dog LOC garden in (LOC = locational verb, it expresses the location of something)
The dog is in the garden. The dogs are in the garden.

有 狗 在 花園 裡
yǒu gǒu zài huāyuán li
EXS dog LOC garden in (EXS = existential verb)
A dog is in the garden. Some dogs are in the garden.
There's a dog in the garden. There's dogs in the garden.

Hmm... Although now that I think about it, I don't know how existentials with definite subjects go i.e. "(then) there's the cat on the roof".

[TomHChappell]

The head of a PP is a preposition, right?

"Head" is actually used in more than one way. Here are two of the most important ones.

(Let's go generativist just for this post.)

_________________________________________________________________________________________

1.
A part of a phrase, which determines which part-of-speech or phrase-type the rest of the phrase can be, and also determines which part-of-speech the whole phrase acts like or the entire phrase's distributional class, is called the phrase's "head", in some writings. If it's just one word, it's called "the head-word" of that phrase.

If you have a production rule

X --> Y + H

or a production rule

X --> H + Y

which is the only production-rule that has an H on the right-hand side, then the H completely determines both what classes or categories of phrases or words can appear in the rest (Y) of the RHS, and also what class or category of phrase the entire phrase (X) is.

A word in a phrase, which determines which part-of-speech the rest of the phrase can be, and also determines which part-of-speech the whole phrase acts like, is called the phrase's "head word", in some writings.

Ordinarily grammarians don't refer to such an H a head-"word" unless H is a non-branching symbol; that is, it is not the left-hand side of any rule with more than one symbol on the RHS nor with a branching symbol on the RHS (and therefore it can only be eventually replaced by a single word). Some grammarians may sometimes define "head" in such a way that the head of a phrase could be a phrase; but they're much likelier to talk about head-words than phrasal heads.

For your explicit case, if adpositions can only occur in production rules such as

Adjective-likePhrase --> Pronominal-or-NominalPhrase + PostPosition

Adverb-likePhrase --> Pronominal-or-NominalPhrase + PostPosition

or only in rules such as

Adjective-likePhrase --> PrePosition + Pronominal-or-NominalPhrase

Adverb-likePhrase --> PrePosition + Pronominal-or-NominalPhrase

then the adposition (pre- or post-) determines the part-of-speech of the rest of the right-hand-side (it's got to be a noun or nominal-phrase or pronoun or pronominal-phrase) and the distributional privileges of the whole adpositional phrase (depending on the adposition, it can be used either as if it were an adjective or as if it were an adverb).

Sometimes the resulting phrase is referred to by its distrubutional privileges; sometimes by the part-of-speech of its head-word.

Determiners, for instance, only appear in phrases in which the other part of the RHS is noun-like and the entire phrase is also noun-like;

Noun-likePhrase --> NounlikePhrase + Determiner

or

Noun-likePhrase --> Determiner + Noun-likePhrase

These phrases are sometimes called "Determiner Phrases", even though they are used, not like determiners, but rather like nouns.

Often, both the distributional category of the resulting phrase, and of the head-word, are referred to in naming the phrase. So, for instance, you might hear of "ad-nominal adpositional phrases" and "ad-verbal adpositional phrases".


_________________________________________________________________________________________


2.
A part of a phrase, which, if substituted for the whole phrase, would result in a clause equally grammatical and with a closely-related meaning, is sometimes called the heaq of that phrase.



For instance, we might have


NounPhrase --> Adjective + Noun

which has a Noun for its head by Definition 2, but has an Adjective for its head by Definition 1.


NounPhrase --> Quantifier + Noun

which has a Noun for its head by Definition 2, but has a Quantifier for its head by Definition 1.


NounPhrase --> Demonstrative + Noun

which has a Noun for its head by Definition 2, but has a Demonstrative for its head by Definition 1.


_________________________________________________________________________________________


As you can see, "head-initial" and "head-final" are terms of which you can't be sure of the meaning until you become sure of the meaning of "head". That varies from writer to writer.

If they're using definition 1, then does "head-initial" refer only to head-words, or also to the phrasal heads of complex phrases (phrases that can contain more than one sub-phrase)?

If they're using definition 2, then "head-initial" implies the opposite order than definition 1, though it implies it for fewer phrases. In other words, if a grammar is definition-1-head-initial it must be definition-2-head-final, but not necessarily vice-versa; and if a grammar is definition-1-head-final it must be definition-2-head-initial, but not necessarily vice-versa.

Definition-2-head-initial and definition-2-head-final are less restrictive terms than definition-1-head-initial and definition-2-head-final.


_________________________________________________________________________________________


There are also the terms "left-branching" and "right-branching".

A production rule is a possibly-branching rule if its right-hand-side contains more than one symbol, or if its right-hand-side contains a possibly-branching symbol.

A symbol is non-branching if it doesn't appear on the left-hand-side of any possibly-branching rule.

So obviously terminal symbols -- symbols that don't appear on the LHS of any production-rule -- are non-branching.

Meta-terminal or quasi-terminal symbols, which can only be replaced by terminal symbols*, are also non-branching. (* That is, any rule that has a meta-terminal or quasi-terminal symbol on its LHS, has at most one symbol, and has no non-terminal symbol(s), on its RHS).

Any non-terminal symbol that's not non-branching is possibly-branching. It appears as the LHS of some possibly-branching rule. It can be eventually replaced by phrase containing two or more words.

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

Some production rules in a grammar may have, on their RHS, a possibly-branching symbol and also a non-branching symbol.

The simplest and strictest definition of a left-branching grammar is as follows:
In any rule that has a possibly-branching symbol and a non-branching symbol on its RHS, (all of) the possibly-branching symbol(s) must precede (all of) the non-branching symbol(s); and, there is such a production-rule.

The simplest and strictest definition of a right-branching grammar is as follows:
In any rule that has a possibly-branching symbol and a non-branching symbol on its RHS, (all of) the possibly-branching symbol(s) must follow (all of) the non-branching symbol(s); and, there is such a production-rule.

So, for a grammar to be left-branching, it must contain at least one rule of the form
X --> Y + N
and can't contain any rule(s) of the form
X --> N + Y
, where Y is possibly-branching and N is non-branching.

And, for a grammar to be right-branching, it must contain at least one rule of the form
X --> N + Y
and can't contain any rule(s) of the form
X --> Y + N
, where Y is possibly-branching and N is non-branching.

- - - - - - - - - - - - - - - - - - - - -

Some grammarians sometimes only consider production-rules in which the non-terminal symbol on the left-hand-side is (one of) the possibly-branching symbol(s) on the RHS.

For them, for a grammar to be left-branching, it must contain at least one rule of the form
X --> X + N
and can't contain any rule(s) of the form
X --> N + X
, where N is non-branching (and X, obviously, is possibly-branching).

And, for them, for a grammar to be right-branching, it must contain at least one rule of the form
X --> N + X
and can't contain any rule(s) of the form
X --> X + N
, where N is non-branching (and X, obviously, is possibly-branching).


Those grammarians' definitions of "left-branching" and "right-branching" are less restrictive than the simplest and strongest definitions.

- - - - - - - - - - - - - - - - - - - - -

There are some languages that are left-branching except that the order of the phrase
X --> Y + N
can sometimes be re-ordered as N + Y , but only when Y happens to be eventually replaced by a single word; thus, in the production-expansion of Y, no possibly-branching production-rules happen to be actually used in producing that particular utterance.
Some grammarians would still call such a language "left-branching".

And, there are some languages that are right-branching except that the order of the phrase
X --> N + Y
can sometimes be re-ordered as Y + N , but only when Y happens to be replaced by a single word.
As you would expect, these same grammarians would still call such a language "right-branching".


_________________________________________________________________________________________

As you can see, a left-branching language would be definition-2-head-initial and a right-branching language would be definition-2-head-final.

But head-final and head-initial aren't logically related to left-branching and right-branching except in phrases produced by rules such as
X --> X + H
or
X --> H + X
; and even then, maybe, only when H is meta-terminal or quasi-terminal.

So they are related only if the second definition of "head" is meant and only if the weaker definitions of "left-branching" and "right-branching" are used.

Also, as you can see, when you read that someone has said some language's grammar is "left-branching" or "right-branching", you have only a general idea of what they could have meant unless they state it explicitly.
In textbooks they're likely to do so (somewhere! ); in journal articles, not.

[p-glyphs]

The head of a PP is a preposition, right?

For your explicit case, if adpositions can only occur in production rules such as

Adjective-likePhrase --> Pronominal-or-NominalPhrase + PostPosition
Adverb-likePhrase --> Pronominal-or-NominalPhrase + PostPosition

or only in rules such as

Adjective-likePhrase --> PrePosition + Pronominal-or-NominalPhrase
Adverb-likePhrase --> PrePosition + Pronominal-or-NominalPhrase

then the adposition (pre- or post-) determines the part-of-speech of the rest of the right-hand-side (it's got to be a noun or nominal-phrase or pronoun or pronominal-phrase) and the distributional privileges of the whole adpositional phrase (depending on the adposition, it can be used either as if it were an adjective or as if it were an adverb).

Is does this mean that the structure I've described initially is "naturally" non-viable? Or just that the overall structure would have to behave in the same way "higher up the tree"? I'm a little confused about "the rest of the right-hand-side" bit... the postposition would be as far right as it goes.

I like how you put it: adjective-likePhrase/adverb-likePhrase. I was viewing it more like a relative clause.

For Umu, definitiveness is based on the NomialPhrase's relative position to the verb. Because adpositions will function as verbs—for PPs, definitiveness will be assigned internally: based on the NomialPhrase's relative position to the adposition D/B/A verb. This leaves the PP's relative position to the VP's verb free to assign this static dynamic distinction (contrasting with NPs which assigned definiteness here).

I don't know what this means as far as directionality or branching now. As much as it's a pretend language, I want it to work/have some naturalistic underpinnings.

Or wait! dose what you describe have more to do with how the PP interacts with other NP, rather than Verbs? Like the underlined:

this
[(table on) book] fall (floor on)
vs this
[book (table on)] fall (floor on)
to make
'a book on the table fell onto the floor'

Here's something I've not even thought about till now. ... I've got a pretzel in my head.

Thanks for highlighting these distinctions. Suppose I need to figure out what I mean when I say "head".... among numerous other things.

[TomHChappell]

Everything in my last post has to do with the terminology used in describing languages and grammars; not with the likelihood or realism or naturalism of any language or type of language.

Frankly, you could ignore the question of "is this language head-initial or head-final or neither?" and "what is the head of this phrase?" and "is this grammar left-branching or right-branching or neither?", and just design your language as you want; then let your readers worry about those questions, if they wanted to.


-------------------------------------------------------------------------------------


But you said you wanted Umu to be "loosely head-final"; and obviously what that means depends on what "head" means.
And you asked "what is the head in an adpositional phrase?".

Definition 2 of "head" obviously doesn't apply to adpositional phrases, because
Noun-or-Pronoun-or-NounPhrase + Postposition
and
Preposition + Noun-or-Pronoun-or-NounPhrase
never act like nouns or pronouns or noun-phrases and never act like adpositions.

So if an adpositional phrase contains a head-word for that phrase, the definition of "head-word" must be definition 1 rather than definition 2.

In which case the head-word must be the adposition; because:
(1) The fact that one word is an adposition requires that the rest of the phrase be or "act like" a noun (or pronoun or noun-phrase)
and
(2) The whole adpositional phrase "acts as" a modifier; either "as if it were" an adjective (e.g. for the English preposition "of") or "as if it were" an adverb (e.g. for almost all other English prepositions, and for the English postposition "ago"); depending on which adposition is used.

And that's what Definition 1 says a "head word" is.


-------------------------------------------------------------------------------------


All of this typology of "head-initial languages" and "head-final languages" and "left-branching languages" and "right-branching languages", and even "SV languages and VS languages" and "OV languages and VO languages", and all the rest of the order typology, is useful only for grouping a bunch of languages together that have something similar in common in their syntax that isn't shared with any languages not in the group (or bunch).

We're (almost) all (at least mildly) interested in it, but, what you really want to do is to describe your language. If it satisfies the definition of some type, those of us who know what that type means may appreciate you saying so; but only because it will make us quicker to understand your description. If your language doesn't fit into any established type it could still be, for all we know, naturalistic and realistic.


-------------------------------------------------------------------------------------


Is any of that re-assuring?


-------------------------------------------------------------------------------------


Your notion of "Topic" and "Comment" doesn't seem to be the same as what most people mean when they say a language has a "Topic-Comment" structure.

A Topic-Comment sentence might be something like:
"The table, a book is on it."

Topic-prominent languages tend to have sentences of the form "Topic, Comment", where "Topic" is a definite noun-phrase and "Comment" is a complete clause.

"Comment, Topic" sentences tend not to come up much in most languages. When they do come up, it's more likely because the speaker belatedly realizes, after uttering the comment, that the addressee may have had no notion of what the speaker intended his/her comment to be about. But of course there are speakers who are especially prone to "Comment, Topic" sentences in various languages; and there are genres in which "Comment, Topic" sentences are common; so I wouldn't be surprised to learn there are topic-prominent languages in which "Comment, Topic" sentences are more common than "Topic, Comment" sentences.


-------------------------------------------------------------------------------------


Deciding that pre-Verbal Subjects are Topics and post-Verbal Subjects are indefinite (or vice-versa) sounds like a cool thing.
So does deciding that definite Subjects are pre-Verbal and indefinite Subjects are post-Verbal (or vice-versa).
You might also decide something similar about Direct Objects or Primary Objects; that is, that such an Object's position before or after the Verb depended on its Definiteness, or on its Specificity.

If you do that, the majority of natlangs that do anything similar probably put the Topics, and therefore the definite Subjects, before the Verb; and/or put the Presentational NounPhrases, and therefore the indefinite Objects, after the Verb.


-------------------------------------------------------------------------------------


Deciding that Adpositions will be Postpositions if their object Noun-Phrase is indefinite, but Prepositions if their object Noun-Phrase is definite, is also a pretty cool idea IMO. I've got no idea whether any real natural language does such a thing; but that there might be ambipositions (words that can either be prepositions or postpositions, depending) is certainly both naturalistic and realistic.


-------------------------------------------------------------------------------------


Deciding that your language won't have adpositions, because you'll be able to use verbs in your language to express the ideas that some other languages would use adpositions to express, is also likely IMO to be attested in some real natural language.

An adposition, like a case-marking, tells how its object noun-phrase is related either to some other noun-phrase, or to the verb-phrase, or to the clause as a whole.
Quite a few of them are locational and/or directional, and/or tell something about motion.
Quite a few of them tell what grammatical (or syntactic) relation (or function) the noun-phrase has to the verb and the clause (that is, Subject, or Direct or Primary Object, or Indirect or Secondary Object).
Quite a few of them have more than one use.
Quite a few have both a syntactic-function use and a semantic-or-thematic-role use (for oblique arguments or for adjuncts).
Most of them are ad-verbal (they tell how the noun relates to the verb or the whole clause, or they are used to modify a verb or an adjective or an adverb or another adpositional phrase).
Ordinarily one -- usually called the Genitive -- is ad-nominal (it basically "turns a noun into an adjective").
Many languages have more than one ad-nominal case, and/or more than one adposition whose adpositional phrases are used to modify nouns.

There's no reason those purposes can't be accomplished by some other part-of-speech than adpositions; verbs, nouns, adjectives, and/or adverbs can be used for most of them.

Some natlangs have "relators" instead of adpositions; these could easily be replaced by relational verbs, that is, two-participant stative verbs (or even three-participant stative verbs, for things like "between".)

I have in fact read of a natlang -- sorry, don't know where nor when nor by whom nor which language -- which uses a verb roughly meaning "pairs up with" or "accompanies" to express the comitative meaning of English's "with". IIRC its root is related to that language's root for the numeral "two", which, IIRC, is not a noun nor an adjective in that language.

Marc Okrand's conlang Klingon (tlHiNaan) uses nouns rather than adpositions for many locational purposes; Klingon-speakers use a noun meaning "area above" in order to say something English-speakers would gloss using the preposition "over".

That might mean, of course, that you'd need two clauses to say something that could be said in just one clause in a language with adpositions; probably one of the clauses would be subordinate to the other.

The grammatical-relations could be replaced by case-morphology, folded into the nominal morphology. Or you might have the verb agree with more than one of its participants, and agree in several accidents, say definiteness and gender as well as person and number; that would fold the question, of which noun-phrase fills which grammatical relation, into the verbal morphology. (Since you seem to want the word-order to be variable, you probably don't want to do like English does and just say the noun-phrase before the verb is the Subject and the one just after the verb is the Direct Object.)

So, yes, you could do without adpositions; and, yes, you could do it mostly by saying "adpositional ideas" using verbs instead.

-------------------------------------------------------------------------------------


When you said

I'm thinking about using the PPs relative position to the verb to signal .... static vs. dynamic distinctions for spatial predicates.

I'm not sure I understood you. Actually, that's an understatement; I'm rather sure I did not understand you.

You could have some ambipositions that were locative on one side of their object Noun-phrase but directional on the other side of their object Noun-phrase.

Say, "to home" (in your conlang) means "at home" (in English) but "home to" (in your conlang) means "toward home" (in English).

Is that what you meant by "static vs dynamic ... for spatial ..."?

Or did you mean something like:

"to school run" (in conlang) means
"run at the school" (in English);

"school to run" (in conlang) means
"run at a school" (in English);

"run to school" (in conlang) means
"run toward the school" (in English);

"run school to" (in conlang) means
"run toward a school" (in English)

?


----------------------------------------------------------------------------

Is does this mean that the structure I've described initially is "naturally" non-viable?

Oh heck no!
Anything at all is viable if it's not meant to be a human language; and if it is meant to be a human language, then there are theories about what's viable and what isn't, but none of them have as much status and certainty as, for instance, the theory of evolution by natural selection.

Or just that the overall structure would have to behave in the same way "higher up the tree"?

Only if the language's grammar really is thoroughly head-initial, or thoroughly head-final, or thoroughly left-branching, or thoroughly right-branching.

For instance, maybe your grammar consists almost entirely of trios of rules such as the following.
For each (major) part-of-speech (or word-class or lexical category) X, there are three rules:

X^_ --> X + XComplement
X^_ --> X^_ + XAdjunct
XPhrase --> X^_ + XSpecifier

then your grammar would (probably, or almost) be definition-2-head-initial.

I'm a little confused about "the rest of the right-hand-side" bit... the postposition would be as far right as it goes.

"Right-hand side" refers to production-rules.

If you write a generativist or context-free grammar of your conlang, it will consist of production-rules that might look something like the following:
X --> Y + Z
which means "an X may consist of a Y followed by a Z".
For that rule, X is "the left-hand side", because it's on the left-hand side of the arrow,
and for that rule, Y + Z is "the right-hand side", because it's on the right-hand side of the arrow.

"Clause --> Subject + Predicate"
"Predicate --> Verb + Object"
for instance, are rules that allow a language's grammar to generate clauses that look like
[Subect [Verb Object]].
"Subject + Predicate" is the RHS of the first of those rules; "Verb + Object" is the RHS of the second.

If your language allows productions like
"ModifyingPhrase --> Preposition + NounPhrase"
or
"ModifyingPhrase --> NounPhrase + Postposition",
then the adposition both determines:
that "the rest of the right-hand side" -- in this case that is the NounPhrase -- must be a NounPhrase (or as near as dammit like one, say, a pronoun);
and determines:
how the left-hand-side (in this case that is the adpositional phrase) behaves (whether like an adjective or like an adverb).

Does that help?

I like how you put it: adjective-likePhrase/adverb-likePhrase. I was viewing it more like a relative clause.

Well, since, if I understand you correctly, you want Umu to have "verbs doing business as adpositions" instead of actual adpositions, for Umu's case you would indeed have relative clauses (for ad-nominal use) and adjunct clauses (for ad-verbal use) "doing business as" adjective-like and adverb-like (as the case may be) adpositional phrases.

Or wait! dose what you describe have more to do with how the PP interacts with other NP, rather than Verbs?

I don't think so.
Most adpositional phrases say how their object nouns interact with the verb, or with the whole clause, (like the "on floor" phrases in your examples).
Many say how their object nouns interact with some other noun (for instance English's "of" phrases, or the "on table" phrases in your examples).
I hadn't actually considered what the order of an ad-nominally-acting adpositional phrase, relative to the noun it modified (rather than to its constituent object noun), might mean.

Like the underlined:
this
[(table on) book] fall (floor on)
vs this
[book (table on)] fall (floor on)
to make
'a book on the table fell onto the floor'

That would make good sense if you had a rule that everything that modifies a noun -- adjectives, genitive phrases, relative clauses, demonstratives, numerals, quantifiers, determiners, etc. -- follows it if and only if it is definite and precedes it if and only if it is indefinite.

And I don't think such a rule is at all unreasonable.

Now you have four rules:

(1) Nouns participating in a verb go before the verb if they are indefinite; nouns participanting in a verb go after the verb if they are definite

That's why it's
... book ... fall ...
instead of
... fall ... book ...


(2) Ad-verbal adpositional phrases and adjunct clauses describing a state existing before the main clause go before the verb; such phrases or clauses describing circumstances during the event (such as paths or manners or means) or after the event (such as resulting states) go after the verb.

That's why it's
... fall ... [floor on] ...
instead of
... [floor on] ... fall ...


(3) Everything that modifies a noun goes before the noun if and only if the noun is indefinite; everything that modifies a noun goes after the noun if and only if the noun is definite.

That's why it's
...[table on] ... book ...
instead of
... book ... [table on] ...


(4) "Adpositions" are preposed to their object nouns if and only if those object nouns are indefinite; they are postposed to their object nouns if and only if those object nouns are definite.

That's why it's
[table on]
instead of
[on table];

it's also why it's
[floor on]
instead of
[on floor].

------------------------------------------------------------------------

So we'd get:

"A book on a table fell on a floor"
[book [on table]] fell [on floor]

"A book on a table fell on the floor"
[book [on table]] fell [floor on]

"A book on the table fell on a floor"
[book [table on]] fell [on floor]

"A book on the table fell on the floor"
[book [table on]] fell [floor on]

"The book on a floor fell from on a table"
[on table] fell [[on floor] book]

"The book on the floor fell from on a table"
[on table] fell [[floor on] book]

"The book on a floor fell from on the table"
[table on] fell [[on floor] book]

"The book on the floor fell from on the table"
[table on] fell [[floor on] book]


---------------------------


Or did I misunderstand?

[p-glyphs]

Cheers Tom

Is any of that re-assuring?

Yes. In honesty, if you had tried to tell me otherwise I'd have said, "too bad, my language my rules."

Your notion of "Topic" and "Comment" doesn't seem to be the same as what most people mean when they say a language has a "Topic-Comment" structure.

I know what you're saying. Portuguese/French uses stuff like "the elephants, they are big" or "my mother, she speaks well." This isn't what I'm going for. I'm definitely in the topic-prominent camp... just in reverse.

Deciding that pre-Verbal Subjects are Topics and post-Verbal Subjects are indefinite (or vice-versa) sounds like a cool thing. So does deciding that definite Subjects are pre-Verbal and indefinite Subjects are post-Verbal (or vice-versa).

At this point I'd use the term 'argument'. I'm still working this out. I have this idea where "subjects' are victimized by the state or action. I want to de-emphasize the notion of subject as much as possible = requiring sentences to have "victims" but not "perpetrators". http://pseudoglyphs.wordpress.com/2011/ ... and-verbs/ I think this would class the language as syntactically ergative, from my understanding of it. The "victim" concept also means I have a lot of room to play with thing like volition or intransitives/unaccustives. The whole agent/patient/sole-argument business leave alot of wiggle room to throw in something cool.

This is why I want topics = NP and comments = complete clause... to totally untie topic from the VP while still being able to hold a "subjectless" conversation.

I'm not sure I understood you. Actually, that's an understatement; I'm rather sure I did not understand you.

I imagine a spectrum running left to right (when using a left to right writing systm). The close you get to the right (the end of the sentence), the more topical each phrase becomes... ending in the topic. Spacial as in, it's location in space.

So when a locative phrase (often a PP... maybe always) comes before the verb (farther left), it signals the location of the argument as a result of the action... cause the action is more topical.

When the locative phrase come after the verb, it location is more topical so it signals the location where the action takes place.

I don't know how I'll treat this distinction every time... like at/to/towards. I might not always need an adposition. From your "walk to school" examples, you pretty much get it. it's just flipped the other way around.

(3) Everything that modifies a noun goes before the noun if and only if the noun is indefinite; everything that modifies a noun goes after the noun if and only if the noun is definite.

This sounds cool but it's not what I'm looking to do. I think I'll keep the adjective-like phrase ahead of the noun. This is what I think I mean when I say consistently comment-topic. There'll be a mechanism that allows the topic to "eat" or possess it's comment, turning a clause into a phrase (blue is the flower > the blue flower)... that can then be "eaten" by something else.

As cool as this sounds... I think there'll be too much spiraling around the verb. Those examples I gave were more just me thing aloud/on type... since only then had I considered have an adj-like phrase and an adverb-like phrase in

Cheers, Tom. Can't wait to do this again.

[roninbodhisattva]

Your notion of "Topic" and "Comment" doesn't seem to be the same as what most people mean when they say a language has a "Topic-Comment" structure.

I know what you're saying. Portuguese/French uses stuff like "the elephants, they are big" or "my mother, she speaks well." This isn't what I'm going for. I'm definitely in the topic-prominent camp... just in reverse.

What is your definition of topic/comment, anyway?

[p-glyphs]

Your notion of "Topic" and "Comment" doesn't seem to be the same as what most people mean when they say a language has a "Topic-Comment" structure.

I know what you're saying. Portuguese/French uses stuff like "the elephants, they are big" or "my mother, she speaks well." This isn't what I'm going for. I'm definitely in the topic-prominent camp... just in reverse.

What is your definition of topic/comment, anyway?

On the sentence level, the topic is an NP, the comment is an entire clause pertaining to the topic. But I'm also viewing the head (definition 2) of the phrases (NPs PPs VPs etc) as topic of the phrase... with the rest of the phrase acting as the comment. On an even small level... I'd also tread, say a noun and and adjective as the topic and comment respectively.

this is why I asked whether there's any known correlation between topics and the heads of phrases.

[roninbodhisattva]

Are you thinking in your general model that both the phrase heads and the topics of sentences will be moved out of their base positions?

[p-glyphs]

...will be moved out of their base positions?

I'm not understanding. Please, could you explain what this means?

[TomHChappell]

By "Perpetrators" and "Victims" do you perchance mean "Agents" and "Patients", or "Actors" and "Undergoers"?

I recommend you read http://www.sil.org/linguistics/Glossary ... sticTerms/. It's not gospel but it's pretty good. There's also a French version. I don't know if there's a Spanish or Italian or German or Russian or Dutch or whatever version.

----------------------------------------------------------

In most natlangs, if an utterance has a Topic, then, other things being equal, the Topic comes early.
Topics are always specific/referential, and usually definite as well.

In most natlangs, if an utterance has a Presented element (viz. a new Topic), then, other things being equal, the Presented element comes late.
Presented elemants are usually specific/referential, but often indefinite.

In most natlangs, if an utterance has a Focus-of-emphasis, then, other things being equal, the Focus-of-emphasis comes late.

In most languages, if an utterance has a Ground and a Kontrast, the Ground comes early and the Kontrast comes late.

In most languages, if an utterance has a Given part and a New part, the Given part comes early and the New part comes late.

If the Topic and the Focus are both "sharp" -- say, single words -- then what is neither Topic nor Focus may be called "transitional". If the Topic and the Focus are both "fuzzy" -- say, big parts of the clause -- then what is both Topic and Focus may be called "transitional".

If the above confuses you, but not so much that you can't go look it up, then I've probably succeeded in communicating.

[p-glyphs]

By "Perpetrators" and "Victims" do you perchance mean "Agents" and "Patients", or "Actors" and "Undergoers"?

I've been hesitant to answer that question because I fear it will limit the possibilities of the new language. At first I thought actor/patient was the way to go but I grew more uncertain with time.

like the verb "sleep" for example: the victim of sleep could be the person who falls asleep, the perpetrator could be the person who goes to bed. This is pretty typical, but there are all sorts of ways I can play around with "perpetrators" of intransitive verbs or the "victimes" of transitive verbs... ways that have nothing to do with actual transitivity. I was interested to read about "unaccusitive verbs" recently... which has me thinking down another path entirely.

In most natlangs. [...] If the above confuses you, but not so much that you can't go look it up, then I've probably succeeded in communicating.

I have a pretty good surface understanding of these concepts—though I've been meaning to learn more about "focus" and I hadn't even heard of "ground"—so thanks for this resource. it's been really cool.

That's why I like when people comment on my blog. The engagement forces me to structure my thoughts on paper in a way that makes them understandable to others and thereby more accessible to even myself. I really appreciate the exchange.

chur!

[TomHChappell]

I've been hesitant to answer that question because I fear it will limit the possibilities of the new language. At first I thought actor/patient was the way to go but I grew more uncertain with time.

I think that's why Foley and Van Valin and the other Role & Reference Grammarians say "Actor" instead of "Agent" and "Undergoer" instead of "Patient". If I recall correctly, and if I understood correctly, they think of "Actor" and "Undergoer" as "macro-roles" and/or "proto-roles" rather than as "roles".

like the verb "sleep" for example: the victim of sleep could be the person who falls asleep, the perpetrator could be the person who goes to bed.

Languages with the "Fluid-S" type of Split-Intransitive Morphosyntactic Alignment often use the difference between, say, "I-AGT sleep" vs "I-PAT sleep" to indicate whether I chose to go to sleep and deliberately arranged to do so, or instead fell to sleep whether I wanted to or not, perhaps accidentally and perhaps without realizing it was about to happen.

Some linguists have come up with a hypothesis about some intransitive clauses being "unaccusative" while others are "unergative". "Unaccusative" clauses' Sole participants (or Subjects) are more like Patients; "unergative" clauses' Sole participants (or Subjects) are more like Agents.

This is pretty typical, but there are all sorts of ways I can play around with "perpetrators" of intransitive verbs or the "victimes" of transitive verbs... ways that have nothing to do with actual transitivity. I was interested to read about "unaccusitive verbs" recently... which has me thinking down another path entirely.

And I see you've already heard of those.

I have a pretty good surface understanding of these concepts—though I've been meaning to learn more about "focus" and I hadn't even heard of "ground"—so thanks for this resource. it's been really cool.

I had to look around a lot to find what I've seen about it; and I never kept a bibliography.
You might want to look up Discourse Structure, Information Packaging, and Information Structure.
You might want to look up Pragmatics.
There's another pair or trio of contrasting terms like these, one of which, if I remember correctly, is "tail", but I've forgotten the other one (or two).
I think if you just look up Topic and Focus and Given and New and Kontrast and Ground and the other terms I mentioned, both in that last post and in this one, you'll probably find everything I've ever found. It'll be fun.

That's why I like when people comment on my blog. The engagement forces me to structure my thoughts on paper in a way that makes them understandable to others and thereby more accessible to even myself. I really appreciate the exchange.
chur!

Thanks for saying so!

___________________________________________________________________________________________

next comment not serious

BTW:

Pre/Post-Positions: Where's the head?

If you need to PP, it's the first door on the right down that hallway.