Word Categories

One of the first things that people noticed when they started thinking about language as language was that words tend to fall into categories and that the members of these categories behave in similar ways. The traditional name for those categories is the "parts of speech." In this chapter, we will look at these word categories and see how the traditional account is somewhat misleading, as well as inaccurate. With a more accurate idea of word categories, we will be equipped with the basics that we need to begin studying sentence structure.

The Traditional View: Parts of Speech

You may have forgotten much of the grammar you were taught in school, if you were taught any at all, but most people can remember the parts of speech, at least the major ones. What is a noun? You probably said "a noun is a person, place, or thing." A verb? It describes an action, right? What about a preposition? You may have had more difficulty here, but perhaps you learned that prepositions tell you what an airplane can do to a cloud (go through, under, into, etc.). All of these definitions are well-entrenched in our educational system, but linguists are happy with none of them. If we scrutinize them, the traditional parts of speech turn out to be problematic. Consider the traditional definitions of noun and verb:
Noun: A noun is a person, place or thing. Verb: A verb describes an action or state of being.
These definitions cover what we might call prototypical cases. Nouns often do label objects in the real world (car, tree, apple, etc.) and verbs most commonly express action (run, play, eat, etc.). But what do we do with abstract nouns like love or destruction. One easy way out is to add "idea" to the definition, but this change comes at a severe cost, for "idea" can be taken to encompass just about everything. Consider sentences such as
(1) John gave him a shove. (2) John shoved him.
What allows us to say that shove in sentence (1) functions as a noun, but shoved in sentence (2) functions as a verb? The meaning of both sentences, after all, is essentially the same. And how do we account for verbs like hear or undergo? In a sentence like
(3) Vivica underwent a tonsillectomy as a child.
the subject does not really perform an action, nor does the verb describe a mere state of being. It actually describes a change of state. If we broaden our definition to say that a verb tells us something about some person or thing, it becomes difficult to explain the difference between verbs and adjectives. The traditional definitions of parts of speech founder because they look for semantic definitions. These definitions may cover the canonical situations acceptably, but any definition that covers all cases becomes so vague as to be useless for making discriminations. Another problem with the common way of presenting parts of speech stems from their origins in Latin grammar. The term part of speech, and most of the labels themselves, were borrowed from the study of Latin.[1] When English was first subjected to grammatical analysis, Latin was the language of educated Europeans, and it was presumed to represent an ideal, logical grammar. Therefore the earliest writers of English grammar books simply applied the terminology and classification they knew from Latin to the description of English. Because the two languages have significant grammatical differences, however, the fit was not perfect. Most Latin grammars described eight parts of speech: nouns, pronouns, verbs, adjectives, adverbs, prepositions, conjunctions, and interjections. If one didn't look too closely at the details, these categories worked, more or less, for English. But there were many problematic cases that troubled grammarians from the start. How, for example, should one handle the word the, or the word to when it appears in front of a verb? Latin had no direct equivalent to either word, but some grammarians tried to force these words to fit the Latin categories anyway. Therefore the was considered an adjective and to was called a preposition. Other grammarians disagreed, creating new categories for these words. This disagreement was never resolved in traditional grammar, and to this day, different textbooks make conflicting statements about these words.[2] --- Notes [1] "Part of speech" is a literal translation of the Latin pars orationis. [2] Today, the reason some textbooks differ is likely that they have been influenced by more recent linguistic grammars, but even in the nineteenth century there was never perfect consensus. See, for example, Goold Brown, The Grammar of English Grammars, 6th ed. (1862), who argues for ten parts of speech. This lack of consensus is worthy of note because some textbooks confidently speak of eight parts of speech as if the whole issue had been settled centuries before.

The Linguistic View

Given all this confusion over the concept of parts of speech, it's reasonable to ask if we can't just jettison the concept completely. Why do we really need to know what a noun is? In fact, the problems with traditional parts of speech have prompted some linguists to abandon the term part of speech completely. They have not, however, given up on the idea behind the label. The term part of speech simply means "a word category." In other words, it reflects the important observation that words can be grouped into categories because they behave similarly. For example, consider how we can complete the following sentence frames:

(4) She has no ____.
(5) She can ____.

The first sentence can be completed with words like bicycle, shoes, worries, ability, home, etc., that is, with nouns, but not with words like went, happy, in, or cheerfully (verb, adjective, preposition, and adverb respectively). The second sentence works with words such as hide, fly, delay, lie, cry, etc. (verbs), but not shoes, beautiful, happily, into, etc. Such sentence frames show that there's more to a word than its meaning. Words also belong to categories, and knowing membership in a particular category lets us predict where the word can fit in the sentence.

(6) *My sons both graduation high school.

A sentence like (6) is ungrammatical because the slot that graduation occupies in the sentence requires a verb, not because the general meaning of graduation is inaccurate.

Some linguists avoid the term parts of speech and prefer to speak simply of categories. What is gained by changing the terminology?

It is true that "parts of speech" is misleading if we take the expression literally, as components of language. Clearly, there are many more parts to language than word categories. On the other hand, "part of speech", as a term of art, differs little in its basic meaning from category.

It's really the implication of the term--its association with old grammar books--that causes some to avoid it. I, however, find it hard to see enough difference between the two terms to justify abandoning so familiar a term as "part of speech." Although traditional definitions are muddled, in practical terms, older grammarians meant largely the same thing as modern linguists do with major categories such as noun, verb, or adjective. Even where old fashioned grammarians could not explain the parts of speech adequately, they would still assign the majority of words to the same categories linguists do. (The exceptions, we will see shortly.) In other words, even if traditional grammarians did not define what they were doing very well, their intuitions about these categories led them to many of the same conclusions. So the lexical categories are essentially the same thing as the parts of speech. The fact that the details differ doesn't really affect that essential similarity.

The insistence upon the generic term category, however, does have the virtue of emphasizing just what the parts of speech are, something that is opaque in the traditional term. For that reason, we will use part of speech and category interchangeably, keeping in mind that using the traditional term does not imply we accept the details of traditional classification uncritically. Instead, we will examine how these categories can be redefined to better reflect the way they actually work.


Testing Category Membership

If we are going to do more than simply accept the traditional parts of speech uncritically, we need to establish some sort of theory of word categories, a set of principles that will let us decide where the traditional categories work and where they need revision. Armed with this procedure, we will find that traditional grammars describe some categories that have no real existence in Present-day English. They also conflate other categories which are actually distinct.

Our basic procedure will be to look for elements that are grammatically distinct in English. In other words, we must find structural reasons to distinguish one item from another. For example, we can justify distinguishing verbs from nouns based on the relationships they enter into:

(7) Brown should denounce the need to memorize grammatical definitions.
(8) Brown's denunciation of the need to memorize grammatical definitions.

In example (7), denounce belongs to a category (verb) that can take an -ed inflection to indicate past time (for example, "Brown denounced it."). It can also follow an auxiliary verb (in this instance, should). It can also, in turn, be followed by a noun phrase (the need) that functions as something called the direct object. (Don't worry if some of these terms are unfamiliar. We will cover them in the upcoming chapters.) One way to speak about these possibilities is to say that denounce can enter into a variety of structural relationships with other elements in a sentence. These relationships are not a matter of the word's meaning. Notice that a wide variety of different words can replace denounce. If we were to substitute them, the sentence's meaning would change entirely. Yet all those words appear in the same structural contexts.

The word denunciation in (8) enters into an entirely different series of relationships, even though its meaning is quite similar to that of denounce. It can be preceded by a definite article (the) or a noun phrase marked with the so-called "possessive" ('s), it can take a plural -s inflection, and it can be modified by an adjective (for example, "Brown's quick denunciation). If we try to make denunciation fit into any of the patterns that work for denounce, we get ungrammatical nonsense:

(8a) *Brown's denunciationed of the need to memorize grammatical definitions.
(8b) *Brown's can denunciation of the need to memorize grammatical definitions.
(8c) *Brown's denunciation the need to memorize grammatical definitions.

As a result, we say that denunciation belongs to a different category (noun).

We will use both procedures repeatedly both to explain how we arrive at our categories and to figure out which category any particular word belongs to.

Another important point about word categories is that they exist within a hierarchy. That is, we will recognize both primary categories and subcategories. For example, we accept the primary category of noun, but not all nouns behave the same way. Words like Gina and car are both nouns and share properties such as the ability to appear as the principal word in a subject. But they also differ in the words that appear with them. Car, as long as it is singular, must appear with a word like the or a. Gina, on the other hand, cannot appear with these words:

(9) *Car is in the driveway.
(10) *The Gina was late for work.

We therefore say that car and Gina belong to the same primary category, but different subcategories.

Primary Categories

Contemporary linguistics describes some word categories differently from traditional grammar books, and introduces several new distinctions.

One distinction that is sometimes made is between lexical categories and functional categories.
Lexical categories contain the content words--nouns, verbs, etc. These are the words that carry the primary meaning of the sentence. There are also words that carry little specific meaning of their own. Their main purpose is to serve as the glue to hold the content words together. Such words belong to functional categories. Although this distinction is conceptually useful, it's not always easy to assign categories clearly to one group or the other. Prepositions, as we will see, have some lexical qualities and some functional qualities.

For that reason, we will not make too much of the lexical vs. functional distinction. Instead, we will simply describe the primary categories. We will examine how these categories work in more detail as we learn more about sentence structure. For now, here's a brief overview.


Although I have already tried to show why the traditional definition of a noun (person, place, or thing) is inadequate, now that we have come to define what nouns are, I am going to start with that definition anyway. Am I contradicting myself? Not really. Nouns do refer to people, places, and things, but that doesn't exhaust the extent of their reference. People, places, and things are prototypical nouns. If we're studying a new language, the category that we will call "noun" in that language will be the one that includes these core objects.[1] We will start with these core nouns, observe the patterns that they exhibit, and then use those patterns as a structural test for other words whose category membership may be less clear.

Let's begin with a few examples of such core nouns: teacher, house, car.

All of these words use the same suffixes. They change form to distinguish singular from plural by adding -s:

One teacher, two teachers
One house, two houses
One car, two cars

They also take a different suffix that is traditionally called the possessive ('s for singular nouns, s' for plural ones) although for reasons we will come to later, it's more accurately called the genitive.

the teacher's credential
the house's roof
the car's engine

Nouns can also be formed from preexisting verbs, adjectives, other nouns by adding certain suffixes, e.g., -ment, -tion, -hood, etc. So the presence of such a suffix is good evidence that the word you are looking at is a noun. (See the section on morphology for more information on this process.)

These morphological tests work for a wide variety of nouns, but not all. For example, there are some nouns that form the plural irregularly (e.g., mouse/mice), or show no difference in form at all (e.g., sheep, deer, etc.). Nevertheless, we still want to assign these words to the same category because in other respects they behave just like the more regular words.

Another set of tests looks at the context in which a word can appear in phrases or sentences. As was indicated above, nouns can appear in sentence structures such as the frame in (4), repeated here for convenience:

(4) She has no _____.

Nouns can also appear as the subjects of sentences:

(11) Deprivation is growing among the unemployed.

Nouns also follow certain function words known as determiners (see below), such as the, a(n), my, that, etc. Thus we can say the enrollment, but not *the enroll.


[1] Although the claim is not entirely uncontroversial, most linguists believe that every human language distinguishes at a minimum between nouns and verbs.


Pronouns are words like he, she, or you that let us cross-reference another entity somewhere else in the discourse or in the real world. Traditional grammars state that pronouns replace nouns, but it would be more accurate to say that they replace noun phrases.

(12a) [The airplane parked on the tarmac] appeared damaged.
(12b) It appeared damaged.

In (12b), the pronoun it does not replace just the word airplane of (12a); it replaces the entire string of words, the airplane parked on the tarmac. Replacing only airplane with a pronoun yields an ungrammatical sentence:

(12c) *The it parked on the tarmac appeared damaged.

Pronouns serve the same functions in a sentence that nouns do, most notably they are the heads of noun phrases. They largely observe the same syntactic rules as nouns, for example subject-verb agreement. For these reasons, we will consider pronouns to be a special type of noun rather than an independent word category.

We will use the term referent for the entity to which the pronoun refers. The referent does not necessarily have to be named linguistically. For example, if you and I are standing on a street corner and observe an automobile weaving in and out of traffic at a high rate of speed, you might say to me, "He's driving recklessly." The context of the situation tells me that the referent for he is the car's driver without your needing to use that noun phrase. However, pronouns often do refer to other noun phrases, and in this common situation those noun phrases are called antecedents.

Sometimes, we will need to note what pronoun refers to what antecedent. In this case, we will use a subscript notation. For example:

(13) Genevieve helped Albertj with hisj physics homework.

In (13), the letter j indicates that the pronoun his refers to Albert. In other words, j serves as a co-referencing variable. We can use such subscripts to make assertions about particular interpretations of pronouns. For example:

(14) *Genevievej made herj a sandwich.

We mark (14) as ungrammatical not because it has no sensible interpretation but because her cannot be understood to apply to Genevieve. If her referred to any female other than Genevieve, the sentence would be acceptable.

Pronouns come in several varieties:

Personal pronouns
I, you, he, she, etc. usually refer to a previously mentioned noun phrase or to a clearly implied person.
Reflexive pronouns
myself, yourself, themselves, etc. most commonly refer to the subject of the clause they are in.

(14) The graduating seniorsj threw themselvesj a party.

Because of this requirement that reflexives refer to the subject, reflexive pronouns usually cannot appear in subject position

(15) *Himself went to the party.

For the same reason, transitive verbs with reflexives in the direct object cannot be made passive:

(16a) Ron Howard cast himself in his own movie
(16b) *Himself was cast by Ron Howard in his own movie.

Indefinite pronouns:
somebody, anyone, everything, nothing, etc. don't refer to specific nouns.
Interrogative pronouns:
what, who, or whom, replace a noun phrase in forming a question.
Relative pronouns:
who, whom, which, whose, replace a noun phrase in a relative clause.

Interrogative and relative pronouns occur as parts of more complex structures, which we will study in a later chapter.



In terms of their distribution, verbs are words that can appear after auxiliaries. In the frame sentence (5), repeated for convenience, can is the auxiliary:

(5) She can ____.

We will have more to say about auxiliaries later. For now, we can simply note that they are words like can, could, will, would, shall, should, may, might, or must. In most grammar books, auxiliaries are considered a special type of verb, and that is how we will treat them. But it's important to note that auxiliaries do not behave like most other verbs. In particular, they fail most of the tests for verb-hood given here. For example, the frame sentence (5) cannot be filled in with another auxiliary.

(12) *She can might.[1]

That fact has led some linguists to treat auxiliaries as a separate word category. We will explore the logic for keeping them as a subclass of verbs when we examine the structure of verb phrases.

Morphologically, verbs change form to distinguish tense, and, in the present tense, the third-person-singular from other persons and numbers. Thus we contrast They walk, the present tense, from They walked, the preterite (past tense), and He/she/it walks from I/you/we/they walk. Verbs also take the suffix –ing.

Note, however, that these morphological tests don't work for every verb. Just as there are some exceptions as to how nouns form the plural, there are some exceptions to how verbs form the preterite. Almost every verb does allow -ing to be added, but there are one or two odd cases, such as beware.


[1] In some regional varieties, of English, for example in North Carolina, two auxiliary verbs actually can appear together in the so-called double-modal construction, e.g., "I might could loan you the money." Such sentences, though, are ungrammatical for all the standard varieties of English.



Adjectives typically specify characteristics of nouns, or they limit the application, as in "the large refund," "an enthusiastic participant," or "purple prose." Most often they appear before a noun, although they can also appear in their own phrases after certain verbs known as linking verbs, as in "Wilma looks cheerful." or "They were happy."

Morphologically, most adjectives are gradable. That is, they express the grammatical category known as degree. The basic form of the adjective, which expresses a quality, is known as the positive degree. To express a greater intensity of one of two items, the comparative degree is used, either by adding the suffix –er or with the word more and the basic adjective. To express the greatest intensity among three or more items, the superlative degree is used, either with –est or most.

Gradable adjectives can be tested by adding the word very in front of them. Thus

(13) She is very slow
(14) *Very fools waste time.
(15) *He very adores her.

Some adjectives, however, describe an all-or-nothing state, and aren't gradable. The very test sounds rather odd with these words, as in

(16) ?They were very present at the assembly.

In such cases, the very test won't help us decide whether present is an adjective. Notice, however, that present does pass the other structural tests for an adjective given above. For example, it can appear after a linking verb like were:

(16b) They were present at the assembly.



In traditional grammar, adverb was a catch-all category for everything that was difficult to analyze. Unfortunately, this had the effect of making the category heterogeneous. Some words that are traditionally called adverbs show very different distributions from other words in the same caategory. In some cases, we will not categorize these words as adverbs at all. We will note such cases as they occur in later chapters. We will begin, however, with the most obvious cases.

Adverbs are characteristically used to modify verbs. That is, they perform the same function for verbs that adjectives do for nouns. And indeed, adjectives and adverbs are often closely related, but they do not appear in the same function:

Modifying Nouns Modifying Verbs
adj. new cars *They new drove.
adv. *a suddenly change It changed suddenly.

Many adverbs can also modify adjectives, and some can also modify words of other categories (except nouns), as well as complete phrases and clauses.

verb modifier: The pedestrian appeared suddenly.
adj. modifier: The suddenly hazardous situation took us by surprise.
clause modifier: Suddenly, the pedestrian stepped into the street.

verb modifier: I almost wrecked the car.
adj. modifier: His confusion was almost comical.
adv. modifier: She almost never misses a meeting.
prepositional phrase modifier: The situation was almost beyond repair.

(Note: if you're having trouble seeing why these adjectives and adverbs are modifying the things that I say they are, you might want to read the chapter on phrase structure, and then return to this section.)

Morphologically, many adverbs are formed from adjectives by adding the suffix –ly. Like adjectives, they are also frequently gradable, and can use the comparative and superlative. The very test also works for adverbs.

(20) She exercises very frequently.



A preposition relates one unit in the sentence to something else in the sentence. Prepositions often express relations of space or time, or they mark various grammatical roles. Words like in, to, over, and through are prepositions. As their name implies, they precede something, usually a noun phrase. The phrase that follows a preposition is called the object of a preposition.

(21) in [the yard]
(22) throughout [the ages]

Prepositions are slightly different from the categories we have already examined. They often have distinct meanings of their own, but many prepositions play a more purely functional. Prepositions form a small, relatively closed set of words. There are only a few hundred prepositions in English, as opposed to tens of thousands of nouns, verbs, adjectives, or adverbs. It's easy to invent new nouns, verbs, adjectives, or adverbs. New prepositions, however, are added to the language only rarely.

Prepositions do not have inflectional endings, so we cannot apply morphological tests to prepositions. However, like adjectives, many prepositions are gradable. These prepositions can be preceded by degree words such as right or straight:

(23) She walked right into the wall.

Not every preposition is gradable, however. Of is a preposition, but it cannot be modified by right/straight.

(24) *The relaxed days right of summer were my favorite.

The ungradable prepositions have what are called grammaticalized uses. In other words, the preposition's meaning is not distinguishable from the grammatical construction in which it occurs. For example, compare the use of by in the following sentences:

(25a) His blind date stood by the fountain.
(26a) The report was completed by a committee of experts.

In (25a), by has an identifiable spatial meaning. This use is not grammaticalized. In (26), however, by has no spatial meaning. Indeed, it's hard to say what independent meaning it has. Its function is grammatical: it specifies the following noun phrase (a committee of experts) as the actor in the sentence. Notice that (25a) is gradable, but (26a) is not:

(25b) His blind date stood right by the fountain.
(26b) *The report was completed right by a committee of experts.

Secondary Categories

The remaining categories are called secondary not because they are unimportant but because they have many fewer members than the primary categories. There are tens of thousands of words in the primary categories (with the exception of prepositions) but only a handful of words in the remaining categories.


Determiners are words that appear before nouns and specify ideas such as definiteness, quantity. Traditional grammar books often lump determiners in with adjectives and pronouns, but we will treat them as a primary category. Determiners play an important role in noun phrases. For now we merely list the most common determiners. We will return to them in more detail when we look at NP structure.


The definite article, the, is used to introduce something that can be identified uniquely within the context of the utterance or of general knowledge. For that reason, the is typically used for "old" information. If I say "bring the chair," I assume you already know which chair I'm talking about.

The indefinite article, a/an is used for situations were the reference is not uniquely identifiable. If I say "bring a chair", I don't have any particular chair in mind.


The demonstratives are this, that, these, and those. Like definite articles, they refer to old information. But they also point to specific things: this book or those children.[1] That "pointing" establishes a relative spatial relationship, which is reflected in the contrast between this/these, used for items that are close to the speaker, and that/those, used for items that are further away from the speaker, relatively speaking.


Many determiners express a notion of quantification. That is they specify how much or how many of the head noun there are. Here's a list of some common quantifiers:

all any both each
either enough every few
fewer less little many
more most much neither
no none several some
sufficient what whatever which



One kind of determiner that deserves separate attention is the numeral.[2] Numerals appear in one of two forms: cardinal (one, two, three, etc.) and ordinal (first, second, third, etc.). When numerals appear in front of a noun in order to quantify it (two birds, four cats, etc.) they are determiners. Numerals can also appear as independent nouns in their own right. We will return to this point when we examine the structure of noun phrases.


[1] The technical term for this pointing function is deixis.

[2] We use the term "numeral" in order to distinguish from linguistic number (singular/plural).

Coordinators and Subordinators

Traditional grammars typically have a category called the conjunction and distinguish between coordinating conjunctions and subordinating conjunctions. In point of fact, these two classes of words do not behave the same way at all, and so there is no good reason to think they are subtypes of a larger category. For that reason, we will treat these words as belonging to separate categories.


Coordinators are words that join grammatically equal units together. The principal coordinators are and, but, or, and nor.


Words whose function is to establish an unequal grammatical relationship, (e.g., that, for, to, whether, if).

(27) She asked me whether it was raining

Most subordinators can also function as other parts of speech: for and to can be prepositions, that can be a determiner, etc.), and so we will return to look at subordinators, and how to dintinguish them from other parts of speech, more closely in later chapters.



Interjections are words like oh, hey, ouch, or aha. They stand apart from other parts of speech in that they do not combine with other words in larger syntactic structures. Their primary function is to express feeling rather than to make a proposition about something. Some words—particularly curses like damn—are primarily verbs but can function as interjections: (28) Damn, I'm late for work again.