A language without raising?

[vegfarandi]

I’ve been doing a lot of reworking on my conlang Imutan and one issue that’s come up is whether it is naturalistic for it to disallow raising entirely or mostly.

Imutan is a strict V1 language with a tendency towards VSO.

The way I’ve always pictured raising verbs simply like this:

kigja edelei-muoda progasôm
appear become-INF=high price
prices appear to be increasing

As opposed to:

*kigja progasôm eledei-muoda

I don’t know any language which does this but as I’m reading the Syntax Construction Kit this came to my attention. It seems plausible to me given the V1 structure but I’m no sure which is why I ask.

Also curious if anyone foresees this having further reaching consequences.

[akam chinjir]

If it's V1, a raised subject is still going to end up after the verb, so I don't really see that being an issue (but maybe I've misunderstood).

At one point I thought Akiatu wouldn't have raising verbs; my most recent thoughts about this are here. The main issue I ended up having with the idea is the resulting disconnect between syntax and topicality: syntactically, you end up with the equivalent of it seems prices are rising, even if it's the prices that are topical; it stopped making sense to me that this topical argument couldn't raise to be the matrix subject.

kigja edelei-muoda progasôm
appear become-INF=high price
prices appear to be increasing

One thing: you've glossed the subordinate verb as an infinitive, which implies that syntactically speaking it can't host a subject. That's actually an important syntactic motivation for raising, to provide a syntactic host for a subject that, semantically speaking, belongs in a subordinate clause.

[zompist]

Your example looks fine to me, but I don't know any VSO languages. Linguoboy or Dewrad...?

Topicalization is definitely something to think about. If you have sentences like "John seems to be unlikely to be chosen", does it make a difference if we're already talking about John?

[vegfarandi]

One thing: you've glossed the subordinate verb as an infinitive, which implies that syntactically speaking it can't host a subject. That's actually an important syntactic motivation for raising, to provide a syntactic host for a subject that, semantically speaking, belongs in a subordinate clause.

Imutan’s infinitive can, I use infinitive to refer to a noun-like complementized verb. (Same as Turkish usage).

[akam chinjir]

Imutan’s infinitive can, I use infinitive to refer to a noun-like complementized verb. (Same as Turkish usage).

Ah, then at least you shouldn't have that sort of reason to allow raising.

[vegfarandi]

Your example looks fine to me, but I don't know any VSO languages. Linguoboy or Dewrad...?

Topicalization is definitely something to think about. If you have sentences like "John seems to be unlikely to be chosen", does it make a difference if we're already talking about John?

Topicalization is handled by a verb, eki ‘concern, be about’ so maybe it would simply be:

ek progasôm, kigja edelei-muoda tuo
concern price, appear become-INF=high 3S
as for prices, it appears they are rising

But it wouldn’t be raising. Eh?

[cedh]

I don't think that would count as raising, no. It looks like two clauses in apposition to each other, the first of them functioning as a topic for the second one. (At least if "concern" is really a finite verb in that example.)

Your basic, single-clause example also doesn't really look like raising to me, because the subject of the second verb is in the expected position for a non-raised structure. What exactly is the relationship between the matrix clause and the embedded clause though? Is the embedded clause syntactically an argument of the matrix verb "appear"? And if yes, subject or object? Does the language have similar constructions with other matrix verbs (which might have a different argument structure)? And superficially unrelated but possibly relevant: How are secondary predicates handled?

[akam chinjir]

I happened upon an example of raising from Tongan, which is verb-initial, maybe it's interesting.

ʻe  lava        [ʻo   ako   ʻe  Pita  ʻa  e   lea      faka-tonga]
TNS possible/can COMP learn ERG Peter ABS DET language Tongan
Peter can learn Tongan

ʻe  lava         ʻe  Pita  [ʻo   ako ʻe Pita ʻa  e   lea      faka-tonga]
TNS possible/can ERG Peter  COMP learn       ABS DET language Tongan
Peter can learn Tongan

(Source: Polinsky and Potsdam, Austronesian syntax, 43.)

[Frislander]

I can imagine certain kinds of polysynthetic languages (e.g. Greenlandic, Nootka and the like) might lack raising as it is traditionally thought of, because they express the equivalent of raising verbs in English with suffixes to the main verb. And even then there might also be plenty of languages which just use parataxis like Pirahã seems to.

[vegfarandi]

I don't think that would count as raising, no. It looks like two clauses in apposition to each other, the first of them functioning as a topic for the second one. (At least if "concern" is really a finite verb in that example.)

Right, well, ek here is a kind of coverb, actually. More of a transposed adverbial, really.

Your basic, single-clause example also doesn't really look like raising to me, because the subject of the second verb is in the expected position for a non-raised structure. What exactly is the relationship between the matrix clause and the embedded clause though? Is the embedded clause syntactically an argument of the matrix verb "appear"? And if yes, subject or object?

In terms of Chomskyan grammar, I'm not sure. I've been reading a bunch of articles on how to analyze V-initial languages, and they don't seem to be compatible with Chomsky's basic theories without modification. Different linguists have different ideas on what those modifications are. There's a concensus that Irish and Welsh have an underlying SVO structure that's transformed to VSO, which just seems a bit far-fetched. There's a theory that Tagalog is verb-initial to satisfy prosodic requirements. All of this seems to ignore the fact that from a logical perspective, verb-first makes a lot of sense. Anyway, just using 2D, non-tree representation, I'd say it was something like:

[kigja [edelei-muoda [progasôm]]]
[appear [become-INF=high [price]]]

I'd say edelei-muoda progasôm is a subject of kigja, and progasôm is the subject of edelei-muoda.

Raising would presumably move progasôm to be the subject of edelei-muoda, downgrading it to object status. So a raised version would be kigja progasôm edelei-muoda, perhaps with a pronoun: kigja progasôm edelei-muoda kja.

But I just have this notion that the language has a preference for pushing NPs rightward and thus does not do raising, and several other transformations that are common in seemingly all languages.

Does the language have similar constructions with other matrix verbs (which might have a different argument structure)?

When an intransitive is relativized:

pare gumerôs ê-dôl asê-ptar
wrote come-PTC DEF=man this-DEF=book
the man who came wrote this book

This is in keeping with the" NPs move to the right" notion. Relativives (i.e. participles) normally go after the head, so this is an actual transformation. But gumerôs ê-dôl acts as the subject of the clause.

If the participle were transitive, this wouldn't occur most of the time, and the RC could even be moved all the way to the end.

Another thing, the participle takes on case, so if the verb took a dative subject (perception, emotion and involuntary movement/position verbs do this), you'd get another case form:

kigiv gumerôsim ê-dôl asê-ptar
saw come-PTC-DAT ê-dôl-DIR asê-ptar-DIR
the man who came saw the book

And superficially unrelated but possibly relevant: How are secondary predicates handled?

Secondary predicates are already exemplified in my original sample sentence; they move all the way up and cliticize to the verb:

selei-mauldṙ
make-beautiful

In a sentence:

sele-mauldṙ ȝes tuo
made-beautiful 1S 3S.M
I made him beautiful

The adjective prevents the pronoun ȝes to cliticize, without a complement in its way it would be -es:

sel-es draunna maundṙ
make-1S thing<PL> beautiful<PL>
I make beautiful things

I happened upon an example of raising from Tongan, which is verb-initial, maybe it's interesting.

ʻe lava [ʻo ako ʻe Pita ʻa e lea faka-tonga]
TNS possible/can COMP learn ERG Peter ABS DET language Tongan
Peter can learn Tongan

ʻe lava ʻe Pita [ʻo ako ʻe Pita ʻa e lea faka-tonga]
TNS possible/can ERG Peter COMP learn ABS DET language Tongan
Peter can learn Tongan

(Source: Polinsky and Potsdam, Austronesian syntax, 43.)

So very much look like it does raising, but on an optional basis.

[zompist]

In terms of Chomskyan grammar, I'm not sure. I've been reading a bunch of articles on how to analyze V-initial languages, and they don't seem to be compatible with Chomsky's basic theories without modification. Different linguists have different ideas on what those modifications are. There's a concensus that Irish and Welsh have an underlying SVO structure that's transformed to VSO, which just seems a bit far-fetched.

Minimalism spends a little too much time creating English-based structures and creating rules that make them look universal. You could go with generative semantics instead, which (at one point) suggested underlying VSO structure...

But more to your point... you might also consider Control statements, like "We persuaded John to dance". The difference from Raising is that the main clause doesn't just take an S as subject, but a normal object NP and an optional S. "We persuaded John" is a valid sentence; cf. "We wanted John to dance" / "*We wanted." So, how do you translate this?

persuaded we dance John

persuaded we John dance

persuaded we John dance John

At the least, these should test your intuitions about moving arguments as far as possible to the right!

[vegfarandi]

In terms of Chomskyan grammar, I'm not sure. I've been reading a bunch of articles on how to analyze V-initial languages, and they don't seem to be compatible with Chomsky's basic theories without modification. Different linguists have different ideas on what those modifications are. There's a concensus that Irish and Welsh have an underlying SVO structure that's transformed to VSO, which just seems a bit far-fetched.

Minimalism spends a little too much time creating English-based structures and creating rules that make them look universal. You could go with generative semantics instead, which (at one point) suggested underlying VSO structure...

But more to your point... you might also consider Control statements, like "We persuaded John to dance". The difference from Raising is that the main clause doesn't just take an S as subject, but a normal object NP and an optional S. "We persuaded John" is a valid sentence; cf. "We wanted John to dance" / "*We wanted." So, how do you translate this?

persuaded we dance John

persuaded we John dance

persuaded we John dance John

At the least, these should test your intuitions about moving arguments as far as possible to the right!

Semantically, control verbs are a bit different in that the John argument is an argument of both words: We persuaded John and the point of our persuasion was that John dance.

So my gut tells me Imutan would allow either way, depending on the speakers intended focus. Is it on John or on That John dance.

Wavod-essê John titei (tuo)
OR
wavod-essê titei John

Haven’t decided if the former requires a pronoun tuo or not.

[vegfarandi]

So with raising-to-object verbs, the subject remains?/moves? to the Vinf clause:

spaudiv titei-es han ê-pôlôdrin
try-PERF dance-INF=1S between DEF=music
I tried to dance to the music

This is even though the two verbs take different subject cases:

spaudiv kigi-er titetô
try-PERF see-INF=1S.DAT dancer-PL
I tried to see the dancers