Syntactic criteria for prepositions
Although definitions of parts of speech sound nice, they really don't capture the generalizations correctly much of the time. So in order to answer the question of whether now is a preposition, we need to determine some syntactic diagnostics for distinguishing prepositions from what they are most often confused with, adverbs.
So we want the class of prepositions to be those words that behave syntactically like other words in the same class, and differently from words in a different class. Crucially, in the cases of adverbs and prepositions we can't just look at the places where they both can occur, since these positions will never be probative. In particular, since prepositional phrases and adverbs can both modify verb phrases, we can't use "modifies a verb phrase" as a useful definition of an adverb.
The first thing we need to get rid of is the traditional notion that a preposition is something that always takes an object. This is part of the reason why lots of prepositions get classified as adverbs, but it really doesn't make a lot of sense. To see why, consider the following prepositions:
- John went in the house. / John went in.
- John ran out of the house. / John ran out.
- John looked down the street. /John looked down.
In each example the same word, with the same meaning, can appear both with an object and without an object. So defining "Preposition" as something that takes an object (making the other versions adverbs) makes very little sense, and doesn't capture the pattern very well. We don't say that an optionally transitive verb becomes something other than a verb when it is used intransitively, so we shouldn't say that an optionally transitive preposition becomes an adverb when used intransitively either; it's just a preposition without an object.
Now when we look at these prepositions, we find that both the transitive and the intransitive ones can be modified by right:
- John went right in the house. / John went right in.
- John ran right out of the house. / John ran right out.
- John looked right down the street. /John looked right down.
Adverbs behave differently from prepositions
This property of being modifiable by right is a property of prepositions in English, and fails when applied to things that are uncontroversially adverbs: (I'm excluding the dialects in which right can modify adjectives and adverbs with a meaning of "very"; in those dialects this test is weaker, obviously.)
- John ran quickly.
- *John ran right quickly.
- John eats apples frequently/often
- *John eats apples right frequently/often.
In addition to being not modifiable by right, adverbs are impossible as the complement of the copula verb be. This distinguishes them from almost all other categories in English. But all of the intransitive prepositions are perfectly fine after the copula.
- *John is frequently/quickly/often.
- John is out/in/outside/inside.
"Now" is a preposition
We now have some tests that we can use to determine whether now behaves more like an intransitive preposition or like an adverb. With both of these tests, now behaves like a preposition and not like an adverb.
- John is leaving right now.
- *John is leaving right immediately.
- The meeting is now.
- *The meeting is immediately.
We can also show that immediately can modify temporal prepositions, while now cannot:
- The meeting is immediately after the game.
- *The meeting is now after the game. (≠ "immediately after the game")
So we have little evidence that now is an adverb. The fact that dictionaries call it an adverb is simply because they call all intransitive prepositions adverbs, but this is not supported linguistically, which is why (almost) all linguists (at least ones who study syntax) will tell you that now is a preposition.
A more nuanced version
Strictly speaking, what we have shown in the discussion above is that now has the same distributional properties as a prepositional phrase, and in this respect is distinct from an adverb phrase. What it doesn't show exactly is that now is a preposition as a word.
Pronouns and nouns
To make sense of this difference, we can consider the difference between pronouns and nouns:
- John met a man/the man/that man.
- John met Bill
- John met him.
Traditionally we say that phrases like a man, the man, that man are noun phrases, because they contain a noun and that seems to be the most 'important' element of them. Now we know that pronouns like him substitute for noun phrases, but we typically don't want to analyze pronouns as nouns. Conversely, even though proper nouns like Bill don't have usually have determiners, we don't want to say they are not nouns. The reason for this is that pronouns cannot themselves take determiners, but proper nouns can:
- *The him/he that is in my stats class is nice.
- The Bill that is in my stats class is nice.
Some syntacticians (including myself) take this fact (along with some others) to be evidence that what we traditionally call noun phrases are in fact determiner phrases, and pronouns are in fact "intransitive" determiners, while proper nouns are nouns with a null determiner. This means that pronouns, proper nouns, and traditional noun phrases with determiners will all have the same phrasal distribution.
There are a couple of simple argument for the view of pronouns as determiners. First, it allows for a simple characterization of the demonstrative determiners: (this may be an oversimplification, but I will leave that aside.)
- I liked that book.
- I liked that.
We can treat that as an optionally "transitive" determiner: in 23 it takes an NP and in 24 it doesn't.
Second, in many languages the definite determiner forms are morphologically identical to the weak pronouns. This is true e.g. in French and Spanish, for example (data from French, but equivalent data can be constructed in Spanish).
Je lis le livre/les livres.
I read the.MSg book/the.MPl books
"I read the book/the books."
Je le/les lis.
I it/them read
"I read it/them."
Additionally, the idea that proper nouns are nouns with null determiners receives more support because in many languages proper nouns show up with a definite article. (e.g., Portuguese, Greek, and also dialects of Spanish and German.)
From this discussion we can conclude that pronouns are not nouns themselves, but are more likely "intransitive" determiners, and this is what allows them to be pro-forms for other Determiner+Noun combinations.
Applying this analysis to Prepositions
Assuming this analysis of noun phrases as DPs is correct, we can apply a similar logic to the analysis of prepositional phrases. We know that there are pro-prepositional phrase elements, like there and then. These elements can also be modified by right (right there, right then) and so pass our test for prepositional phrase.
- John put the book on the shelf/under the table.
- John put the book there.
- John left at noon/on Tuesday.
- John left then.
It's obvious that now is much more like these elements as well, since it is anaphoric to the speech time, just like then is anaphoric to some non-speech time.
- I have to leave at 3:00 PM.
- That's now!
So just like a pronoun substitutes for a Determiner+Noun combination but is not itself a noun, but rather a determiner, we could say that there and then (and therefore now) are pro-prepositions, i.e., an element analogous to the determiner in a determiner phrase. For concreteness, let's call this category R and therefore there and then and now are RPs.
This may well be the correct way to analyze these elements, but it does come at a cost: if we want to say that the pro-forms are not prepositions themselves, but of category R, we are now committed to saying that all the things we thought of as prepositional phrases are RPs with null R elements. Unfortunately we don't seem to find any R-words that combine directly with PPs (i.e., there is no R-element that is analogous to the, for example.
I'm actually quite sympathetic to this kind of analysis on theoretical grounds, but I don't think it gains us very much descriptively, which is why I think that on descriptive grounds, calling there, then and now prepositions captures their syntactic behaviour quite nicely.