What is a preposition?

Traditional grammars define prepositions as always being followed by a noun phrase (a phrase headed by a noun).  However, Huddlestone and Pullum (2005) argue for a broader definition of prepositions. Their wider definition also captures some items traditionally viewed as subordinating conjunctions or adverbs.

Table 1 illustrates some problems caused by the traditional classification. It shows 3 sentences headed by the verb know and 3 more phrases headed by before. In sentence 1, the complement of know and before is a noun phrase. In sentence 2 it is a clause and in sentence 3 know and before have no complement.

Type of complementKnowBefore
Noun phraseWe know the last actWe left before the last act
ClauseI know he diedThat was before he died
NoneI had seen her beforeYes, I know
Table 1. Complements of verbs and complements of prepositions

Everyone agrees that know is a verb in all 3 cases. But even though the 3 instances of before mean the same, traditional grammars analyse them differently:

  • as a preposition in sentence 1
  • as a subordinating conjunction in sentence 2
  • as an adverb in sentence 3.

Moreover, the same modifiers (for example, long, right, shortly, an hour or a short while) can precede before in all 3 instances.

I now look first at the category classified traditionally as subordinating conjunctions and then at the category classified as adverbs.

Subordinating conjunctions

Huddlestone and Pullum identify 3 sub-classes of what are traditionally classified as subordinating conjunctions:

  • items that can also be used as prepositions before a noun. These items include after, because, since, till. Huddlestone and Pullum analyse these items as prepositions that can be used with either a nominal complement (a complement that is a noun phrase) or with a clausal complement (a complement that is another clause).  
  • items such as if [when if has a conditional meaning], although, because, lest, provided, though, unless. Huddlestone and Pullum analyse these items as prepositions that can only be used with a clausal complement.
  • that, whether, if [when if means the same as whether] Huddlestone and Pullum analyse these 3 items as subordinators. They state that no other subordinators exist in English.


Subordinators introduce a subordinate clause and mark that clause as subordinate to the main clause, but without introducing any meaning:

  • that serves only to mark the clause as subordinate. It can sometimes be omitted.
    [I think [(that) she is right]]
  • whether (and its synonym if) also mark the subordinate clause as interrogative—asking an (indirect) question—so whether (and synonym if) cannot be omitted.
    [I wonder [whether she is right]]

In the examples:

  • [brackets] mark the clause boundaries.
  • The subordinate clauses (that) she is right / whether she is right are embedded in the main clauses I think / I wonder.
  • brackets around (that) convey that (that) is optional.

Unlike these 3 subordinators, the other items traditionally classified as subordinating conjunctions have independent meaning and they head the clauses that are their complements. For example, in she stayed after the others left, the word after conveys a meaning of timing. It performs the same function as in she stayed after the departure of the others. In that latter phrase, after the departure of the others is clearly a prepositional phrase headed by the preposition after.

Items traditionally classified as adverbs

This section discusses 2 sub-classes of items traditionally classified as adverbs:

  • items that can optionally take a noun phrase complement
  • items that never take an NP complement

Some people use the term particle for some or all of these items.

Prepositions with an optional NP complement

Some items are traditionally classified as prepositions when they have a noun phrase complement (eg inside the room), but as an adverb if they have no complement (eg inside). Huddlestone and Pullum suggest classifying these items as prepositions in both instances. They give the following arguments for not treating these items as adverbs when they have no NP complement

  • unlike adverbs, they can occur as dependents of both nouns (the temperature inside) and verbs (he sat inside). Although adjectives or prepositional phrases appear as dependants of nouns, adverbs cannot do so: a manager with tact / a tactful manager; but not a manager tactfully
  • unlike adverbs, they can occur as a complement of the copular verb be when used in ascribing a property to the subject of be (she is inside). Huddlestone and Pullum distinguish this ascriptive use of be (Jane was enthusiastic) from its specifying use, which specifies the subject’s identity (the last person to leave was Jane). Adjectives can appear in that ascriptive construction (she is enthusiastic), but adverbs cannot (‘she is enthusiastically’ is impossible).
  • the traditional class of adverbs is very broad. Removing these items makes adverbs a less disparate class. The remaining members of that class would fit better with traditional definitions of an adverb as words that modify a verb, an adjective or another adverb.

Prepositions that never take an NP complement

Some other items traditionally classified as adverbs never have a noun phrase complement, but can be dependants of nouns or be complements of the copular verb be. Some of these items normally occur without complements (eg abroad, downstairs, here, outdoors, overboard, overseas, there). Others (such as instead) have prepositional phrases (not noun phrases) as complements (instead of wine). The following table shows examples of these items acting as dependants of nouns or as complements of be.

Dependent of nounComplement of be
the office downstairsthe chairs are downstairs
water instead of winethis is instead of wine

Because these items can appear in the same place as propositions, Huddlestone and Pullum suggest classifying them too as prepositions.  


Prepositions precede their complements. A similar item that follows its complement is called a postposition. English only has prepositions, with perhaps 1 or 2 minor exceptions. So for simplicity this post uses only that term. A term covering both prepositions and postpositions is adposition.


I find H&P’s arguments convincing. One thing bothers me slightly. They don’t offer a clear definition of a preposition. They offer only the following ‘fairly useful general definition’:

a relatively small category of words, with basic meanings predominantly having to do with relations in time or space and time, containing among its prototypical members grammaticised words that serve to mark various grammatical functions.

Their glossary (in the 1st edition) contains a slightly condensed version of this (there is no glossary in the recent 2nd edition of their textbook):

a category of words whose most prototypical members (in, on, under, before etc) denote relations in time or space and take NPs [noun phrases] as complement (in the car, on the chair).

In their discussion, H& P do mention 2 other properties of prepositions:

  • they occur in a range of functions, ‘notably’ as dependents of either nouns or verbs (including, as a special case, as complements of the copula verb be). This property gets a little closer to defining the essence of a preposition. But ‘range of functions’ is still too vague. And although Huddlestone and Pullum make a great deal of the ability to act as dependents of either nouns or verbs (and complements of be) in pushing out the boundaries of the category of prepositions, it is still not clear whether those are just 2 examples. Nor is it clear whether there is any common theme underlying these 2 examples.
  • they do not inflect, unlike nouns, verbs and adjectives. This property seems more like an observation than a property that could be used in a definition.


Admittedly, Huddlestone and Pullum do say in their ‘rapid overview’ in chapter 2, that categories such as noun, verb and adjective have not only prototypical members that have the full set of properties associated with their category, but also other members that do not have the full set.

I understand this approach to prototypical and non-protypical (peripheral) members, and that this approach might be helpful. But in my view their ‘general definition’ is too vague to provide a useful starting point in identifying the least prototypical members. Take their examples ‘here’ and ‘there’. I can follow the logic that leads them from clear prepositions such as above to here and there. But I don’t see how to get there directly from their general definition. That definition doesn’t contain enough substance to convey a feel for what a preposition actually is.

In fact, I had originally intended to start this post with their definition, but I found it too vague for me to use it up front.  


A Student’s Introduction to English Grammar, by Rodney Huddlestone and Geoffrey K Pullum (2005, 1st edition). The 2nd edition came out recently, but I haven’t consulted it in detail.

Leave a comment

Your email address will not be published.

%d bloggers like this: