A Thing About Words

Posts From the Editors

April 25, 2013

(Not So) Tidy, (Not So) Little Boxes: Finding Parts of Speech

mwu-bookWe often hear from people who want to pursue a career in lexicography. One student I spoke with summed up his qualifications thusly: “I love words, I love etymology, and I’m really good at grammar.”


I paused.


“What’s a noun?” I asked.


“A person, place, or thing.”


I sighed. “Ah, yes, I remember those days.”


Lexicography isn’t just about recording the meanings of words; one of the primary jobs of the lexicographer is also to figure out a word’s part of speech. This sounds much easier than it actually is.


We tend to think of parts of speech as tidy little boxes, and all a lexicographer has to do is pick a word up off his or her desk and flick that word, like a paper airplane, into the right box. That’s not a bad analogy, provided you make the boxes very tiny, place them randomly about a football field away, and give your lexicographer a lilting headwind that will take those paper-airplane words and send them zooming about the field in exactly the wrong direction. It’s not until you have to assign a part of speech to a word that you realize how flexible most English words are.


Take, for instance, the word apple as used in the sentence, “My favorite dessert is apple pie.” Is it an adjective? It seems to be acting like one. But look up apple in your dictionary, and it’s a noun.


When lexicographers have to assign a word a part of speech, they can’t be strict descriptivists who look only at the citations. If that were the case, then apple would be entered both as an adjective and a noun. Instead, they rely on the citations, the work of grammarians before them, and their own sprachgefühl (or feeling for the language).


In the case of apple pie, we use the work of grammarians of the past to lay out the differences between a “true” adjective and a noun that is used like an adjective. A true adjective, for instance, typically meets four criteria. It’s, one, gradable and can do each of the following: take comparative and superlative forms; be used attributively; and function as a subject or object complement. For apple in apple pie, the citations and our sprachgefühl tell us that it’s apple pie (check for the attributive use), but: the pie can be tastier, but not appler (strike one, not gradable); it can be more delicious or the most delicious, but not more apple or most apple (strike two, no comparable or superlative forms); and it may seem delectable but cannot seem apple (strike three, this time on the complement front). So the lexicographer enters it as a noun which can be used attributively, and feels a little smug about having a good grasp of adjectives. Until encountering the adjectives utter and galore.


When you first begin work at Merriam-Webster, you are trained to stop looking at the grammar of words on a macro level (nouns are persons, places, things; verbs describe action or a state of being; adjectives modify nouns; adverbs modify verbs) and begin delving into the microscopic world of English grammar. At that level of magnification, a part of speech which looks like a homogenous, glassy smooth dome of words is actually full of movement, variation, and life. Boundaries are blurry, and categorization is difficult. Adverbs can function as conjunctions; nouns can act like adjectives; some adjectives can take the definite article, like a noun; some verbs don’t describe an action or a state of being.


The traditional eight parts of speech were initially listed in a Greek grammar from the 2nd century BCE, and we’ve kept with this 2,200-year-old system. There are plenty of  linguists who find this system to be artificial and impractical, but old habits are hard to break: English dictionaries have been following this system since there were English dictionaries. So when you are tempted to consider a career in lexicography, be prepared to tangle with our tricky parts of speech. And if that doesn’t give you pause, a gander at the 181-word first definition for adverb in the New Unabridged just might.

Kory Stamper, associate editor, has worked on dozens of Merriam-Webster products. She writes, lectures on English, and is proud to have defined both "bodice ripper" and "apocalypse" for the Unabridged Dictionary. Follow @KoryStamper on Twitter.

For comments and questions not related to the blog please use our customer service contact form.

14 Responses to “(Not So) Tidy, (Not So) Little Boxes: Finding Parts of Speech”

  1. Scott Campisi says:

    I loved this article. And I would love to see examples of verbs that don’t describe action or state of being.

  2. Ron Moss says:

    …. And I’d like to rid the language of the new (I think) practice of adjectivising adjectives, as in ‘crispy’. To my way of thinking, that’s just baby-talk.

  3. Mike Zachary says:

    Excellent job presenting linguistic subtleties in a friendly format.

  4. Wal says:

    English language is in the state of constant changes, and it’s good. That’s why I like it so much.
    It’s the richest language in the world. And who knows? Maybe the most flexible one?

  5. Simon Ellberger says:

    First off, this is a great article and just the kind of essay I would like to see more of in The Blog.
    The content itself revealed an issue about Merriam-Webster’s defining methods that I feel compelled to constructively complain about; so I will. At one point, the discussion focuses on whether or not apple is an adjective in terms like apple pie. The intent of this seems to be to determine if the use of apple in apple pie merits a separate definition in the New Unabridged from its definition as a noun. The answer that the article appears to comes up with is “no,” because apple is being used as an attributive noun as opposed to a “true” adjective. I would disagree with this conclusion. A definition in common use should be entered if it is significantly different from the other definitions already recorded in the New Unabridged, or those older definitions should be changed to accommodate this unrecorded one. The decision should not instead be based solely on how apple is to be classified grammatically, especially when the grammatical system (as implied in the article) being used may very well be an antiquated and impractical one. Using such a system is likely to lead to antiquated and impractical results, which I believe is the case with Merriam-Webster’s handling of attributive nouns.
    The New Unabridged deals with the “definition” of apple as an attributive noun by indicating that apple is a “noun, often attributive” and then giving the first definition of apple as “the pome fruit of any tree of the genus Malus being important economically especially in North America, Europe, and Australasia and markedly variable but usually round in shape and red, yellow, or greenish in color”; it then gives as an example apple dumpling. But this doesn’t tell me what apple means in that attributive sense; I would even argue it misleads me as to its meaning. In the term apple dumpling (or apple pie), if I were to take a stab at defining the use of apple, it would come out initially as “containing the baked peeled portion of an apple as a main ingredient” (a definition Merriam-Webster could doubtlessly improve upon). The point is it is significantly different than the definition of apple as a noun. This might be okay if “often attributive” always meant that the noun being defined is a baked ingredient of something when used attributively, but we know this isn’t so.
    As the article says “old habits are hard to break,” a cliché that seems applicable here to Merriam-Webster’s methodology for defining attributive nouns. The article also says: “When lexicographers have to assign a word a part of speech, they can’t be strict descriptivists who look only at the citations. If that were the case, then apple would be entered both as an adjective and a noun.” And I would say that in a descriptive dictionary, lexicographers should be descriptivists and enter separate definitions for apple as a “standard” noun and as an “attributive” noun. There is no logical reason for the New Unabridged to omit a specific definition for the commonplace attributive use of a noun, especially now that space is no longer an issue. The same is true for adjectival uses of present and past participles of verbs, and for noun uses of gerunds. I think it would be great if Merriam-Webster revised its defining paradigms accordingly, though I realize this to be unlikely.
    None of this, however, takes anything away from Ms. Stamper’s wonderfully written disquisition. As a descriptive endeavor, it was nonpareil. Thanks for providing us with it.

  6. kjc says:

    i agree with Simon.

  7. Grandma Tomato says:

    Why is it that you are not strict descriptivists in this area but you seem to be when defining words?

  8. Ruth Treloar says:

    Perhaps the grammar itself needs another category for such cases when 2 nouns are not joined (i.e., applepie) but used together so that 1 becomes the other’s descriptive.

  9. GL says:

    Is the first criterium proposed for ‘what typically makes up an adjective’ not slightly misleading? While adjectives are more commonly gradable, it isn’t really judicious to assign a ‘strike’ to a word one is weighing as an adjective if it does not meet the first ‘typical’ requirement of being gradable. Then, we would be striking out a host of non-gradable ‘true’ adjectives (e.g. dead, alive, married, single, beautiful, freezing).

  10. Gabriella Mussi says:

    I’m Italian, I love languages and grammar. I’m a Maths teacher and I like rules, theorems, evidence, proof… But I like this flexibility and this indefiniteness too. I wouldn’t like to pursue a career in lexicography, but I’m very interested in the meaning of words and in the structures of language, comparing and contrasting Italian and foreign languages. Thank you for this article.

  11. Marc says:

    To Simon: Your passion for language is admirable, but I think you confuse a dictionary, which is a compendium of word usage, for an exhaustive description of reality. If the special meaning of “apple” in “apple dumpling” merits a dictionary entry as “containing the baked peeled portion of an apple as a main ingredient”, then what of “apple” in “apple fritter”? It’s not baked, and needn’t be peeled. Does this meaning likewise merit special consideration in the dictionary? I think MW’s use of “noun, often attributive” is entirely correct in this case.

  12. WoundedEgo says:

    I’ve heard and tend to believe the theory that adjectives began as nouns in speech where the usage suggested that one modified the other (apple pie is an example). Later, suffixes and other subtlety arose to describe more subtle situations. Makes sense to me.

  13. Simon Ellberger says:

    To Marc: Hi! I appreciate your feedback on my post. Thank you. Concerning your belief that I confuse a dictionary for an exhaustive description of reality, I assure you I have no such misconception. As to your comment about whether the meaning of “apple” as used in “apple fritter” merits special consideration, the answer is, “It depends.” First, Merriam-Webster would have to decide if that meaning meets the criteria for entry. In their article “How does a word get into a Merriam-Webster dictionary?” they state: “To be included in a Merriam-Webster dictionary, a word must be used in a substantial number of citations that come from a wide range of publications over a considerable period of time. Specifically, the word must have enough citations to allow accurate judgments about its establishment, currency, and meaning.” If a definer judges that it meets this criterion, as well as the other criteria Merriam-Webster uses, then this meaning should be covered in the New Unabridged. However, that doesn’t necessarily mean giving it special consideration by adding an additional definition of “apple.” There are at least two other possibilities.

    The first of these options is to modify or expand a definition that already exists so as to include the “apple fritter” usage. As an example, the definition I suggested could be modified to “containing parts of one or more apples as a main ingredient often in baked form,” or in whatever better way a trained lexicographer deems appropriate.

    A second option is to forget entirely about what grammatical category “apple” fits into, and enter “apple fritter” (or “apple pie”) as a single unified “word,” just as “active optics” has been entered into the New Unabridged.