The syntactic structure of PPs in Det function

[BrettRey]

In over thirty dollars, is thirty a fused DetHead? What about at least thirty dollars?

[nschneid]

p. 432 refers to "at least" as a PP within the DP. We can treat "least" as a fused Mod-Head (cf. "the best").

p. 433, "PPs in lieu of DPs" would cover "over thirty dollars". "In lieu of", not "wrapping", suggests that we could just say the number is noun heading an NP.

[BrettRey]

I think least (along with at all, at most, and at once) would be a fused Det-Head, not a modifier of any sort. Only in cases like at the most and at the least would it be a fused Mod-Head.

We could say that the number is an N, but I don't think that's right. For one thing, its number matters for agreement. Consider, also, under approximately 30 dollars, where it's hard to come up with another case where the 30 in approximately 30 is clearly a noun.

That leaves us with the ugliness of a fused Det-Head inside a fused Det-Head in cases like We've got just under 30. There's also cases like under about 30 people, where you have a two Det-Head:PPs.

[nschneid]

Good point about number agreement. Isn't that really an argument against treating "about 30" as a PP in the first place? Because ordinarily it is the thing in Det function that agrees with the head nominal. Odd to say that a PP has number. At best it is a special subtype of PP where the head (preposition) is transparent with respect to morphosyntax.

Would you mind making a tree of what you're thinking for "under about 30 people"?

May be worth pulling in John and Geoff on this one.

[BrettRey]

texstudio_BMOOpX.pdf

\documentclass[tikz,border=12pt]{standalone} \usepackage[linguistics]{forest} \usepackage{times} \usepackage{xcolor} \usepackage[T1]{fontenc} \pagestyle{empty} %---------------------------------------------------------------------- % Node labels in CGEL trees are defined with \Node, % which is defined so that \Node{Abcd}{Xyz} yields % a label with the function Abcd on the top, in small % sanserif font, followed by a colon, and the category % Xyz on the bottom. \newcommand{\Node}[2]{\small\textsf{#1:}\{#2}} % For commonly used functions this is defined with (function) \newcommand{\Head}[1]{\Node{Head}{#1}} \newcommand{\Subj}[1]{\Node{Subj}{#1}} \newcommand{\Comp}[1]{\Node{Comp}{#1}} \newcommand{\Mod}[1]{\Node{Mod}{#1}} \newcommand{\Det}[1]{\Node{Det}{#1}} \newcommand{\PredComp}[1]{\Node{PredComp}{#1}} \newcommand{\Crd}[1]{\Node{Coordinate}{#1}} \newcommand{\Mk}[1]{\Node{Marker}{#1}} \newcommand{\Obj}[1]{\Node{Obj}{#1}} \newcommand{\Sup}[1]{\Node{Supplement}{#1}} %---------------------------------------------------------------------- \begin{document} \begin{forest} where n children=0{% for each terminal node font=\itshape, % italics tier=word % align at the "word" tier (bottom) }{% % no false conditions, so empty }, [NP [\Det{PP} [\Head{P}[under]] [\Obj{NP},s sep=-2em [\phantom{X}\hspace{-4em},tier=dh] [\small\textsf{Determiner-Head:}\PP,no edge,tier=dh [\small\textsf{Head:}\P [\textit{about}] ] [\Obj{NP},s sep=-2em [\phantom{X}\hspace{-4em},tier=dh1] [\small\textsf{Determiner-Head:}\DP,no edge,tier=dh1 [\small\textsf{Head:}\D [\textit{30}] ] ] [\small\textsf{Head:}\Nom [\hspace{-4em}\phantom{X},tier=dh1] ] ] ] [\small\textsf{Head:}\Nom [\hspace{-4em}\phantom{X},tier=dh] ] ] ] [\Head{Nom}[\Head{N}[people]]]] \end{forest} \end{document}

[nschneid]

I don't understand why "about 30" should be Det-Head. Why not "under" taking a PP complement?

[BrettRey]

I don't think under takes about PP complements in general, does it? *I hid it under about the pillows

[BrettRey]

With agreement, it would be similiar to

1. [More/Less/Fewer than twenty] people came to the meeting.
2. [More/Less/Fewer than one] person came to the meeting.

[nschneid]

No, but take doesn't take PP complements in general either, yet it licenses the measurement ones (It took under an hour).

[BrettRey]

I don't think the between argument in the PPs in lieu of DPs section on p. 433 is so clear as they suggest by saying "there can be no question." Consider between the predicted and the measured strains. And the idea that from takes a to PP complement (or that to takes a from PP complement) also seems fishy.

[BrettRey]

The complement with take is an NP. Under is inside the NP somewhere.

[BrettRey]

Actually, I guess my between examples isn't relevant as it can't function as a subject.

[nschneid]

You have "least" in "at least" as an Adj at https://simple.wiktionary.org/wiki/least (but think that may be an error).

"The most", "the least", etc. is discussed at p. 395. Adjectival "least" is discussed on p. 1165, noting that it is the opposite of "greatest". I suppose the opposite of "at least" is "at most" rather than "at greatest", so that could be a reason to call it a determinative.

[BrettRey]

Right! Lesser & least, with no base form. I should trust my past self with Wiktionary.

[BrettRey]

So adjective least should be paraphrasable by smallest or slightest. I don't think at smallest or at slightest works, and so, in analogy with at most, I think at least has a D.

[nschneid]

The situation regarding Det-Head fusion is complicated by the necessity of "the" in certain situations:

• Kim has 15 dollars and Jim has a thousand cents. Who has more money and who has less (money)?
• Kim has 15 dollars, Jim has a thousand cents, and Nim has a dozen bucks. Who has the most money and who has the least (money)?

p. 395: "We analyze this the as modifying the comparative quantifier", so :Det(-Head) [DP :Mod [D the] :Head [D least]].

I suppose calling the a modifier suggests that least might serve as Det-Head even without the.

What do you make of the comparison with other at expressions:

Expression	Adjunct	Det function	Omitted nominal
[i] at most/least	It'll cost 5 dollars at least.	It'll cost [at least 5] dollars.	It'll cost several dollars—at least 5.
[ii] at best/worst	It'll cost 5 dollars at best.	*It'll cost [at best 5] dollars.	?It'll cost several dollars—at best 5.
[iii] at (a) maximum/minimum	It'll cost 5 dollars at minimum.	*It'll cost [at minimum 5] dollars.	*It'll cost several dollars—at minimum 5.

[BrettRey]

This use of the seems fine to me. Small change: :Det(-Head) [DP :Mod [DP :Head [D the]] :Head [D least]]

The comparison with other at expressions also seems to me to point to D.

[nschneid]

How would you analyze "at least an hour"?

[BrettRey]

That pilcrow should be a P

[nschneid]

So if I have it straight, the following structures (leaving off Nom nodes for brevity):

Approximator expression	Approximator + article	Approximator + numeral
approximately	[NP :Mod [AdvP approximately] :Head [an hour]]	[NP :Det [DP :Mod [AdvP approximately] :Head [D 5]] :Head [hours]]
at least	[NP :Mod [PP at :Obj [NP :Det-Head [DP least]]] :Head [an hour]]	[NP :Det [DP :Mod [PP at :Obj [NP :Det-Head [DP least]]] :Head [D 5]] :Head [hours]]
about	[PP about :Obj [an hour]]	[NP :Det [PP about :Obj [NP :Det-Head [DP 5]]] :Head [hours]]

[BrettRey]

Yes, and I have no problems with the first two rows, but I'm bitter about the third, specifically the rightmost cell.

[nschneid]

I'm unsure about the whole thing. :D

I take it the argument that approximators sometimes belong to a DP, rather than as a peripheral modifier, is strongest where the noun is not a unit of measure, like "approximately 5 cookies" (*approximately cookies).

For me, "At the very most twenty people agreed to help" (p. 432) sounds forced. I am inclined to interpret "at the very most" as a supplement to "twenty people", and "at least" as a fixed expression in this usage before a counted noun that is not a unit of measure. But that may be a point of variation.

The idea that "about an hour" would be a PP occurring where an NP would normally be expected definitely bothers me. This is described on p. 646:

By the way, it is not just in object function, but also cases like "about an hour earlier/ago" or "about an hour after we left" or "about a mile away". Apart from approximators, the extent is typically not expressed with a PP.

• They will arrive in (about) an hour, They will arrive after an hour, but *They will arrive in after an hour. I.e., "about" does not pattern like "after" here.

It's as if the PP shell is invisible to the licensing phrase. An analogy can be made to head nouns that are transparent w.r.t. number: "[A bunch [of cookies]] are...". But there, at least we're talking about two NPs, not a PP that behaves externally like an NP.

[nschneid]

They sat on a stool. They sat right on a stool. the stool they sat (right) on the stool on which they sat

They spent a year here. the year they spent here

They spent about a year here. They spent just about a year here. the year they spent (just) about __ here the year about which they spent here

[nschneid]

I guess I would feel more comfortable calling "about" an intransitive PP peripheral modifier within the NP. This construction could be explained as having grammaticalized from reanalysis in a predicative context (easier to see with "over" or "under" than "about"):

The duration of the event exceeded an hour. The duration of the event was [PP over [an hour]]. (duration > an hour; "over" expresses the relation) The duration of the event was [NP over an hour]. (duration = over an hour; "over" is an approximator)

Also, "It took only about an hour"—feels like "only" is a peripheral modifier in the NP headed by "hour".

[BrettRey]

Some more data:

1. About the most exotic thing he did was give blood every few months or so.
2. [The most exotic] or [about the most exotic] thing he did was give blood every few months or so.
3. [One cup of x] or [between two and three cups of y]

[BrettRey]

https://www.cambridge.org/core/journals/journal-of-linguistics/article/abs/prepositional-phrases-as-subjects-and-objects1/DE0F8683907FFB82DBD70B61E9E75CDD

[nschneid]

This came up in treebanking today. Here are trees based on the above analysis, which seems to be what we are currently doing in CGELBank, though I don't love it.

approx.pdf

[BrettRey]

Me neither, for what that's worth.