I'm looking for a maximum subset of the English language such that words can be divided into two categories, one of which must include at least nouns, proper names, adjectives, numerals, verbs and adverbs, and the other must include at least the articles ('a', 'the'). Each of the groups may, and should, include other words, like different kinds of pronouns, etc.

It is OK to omit some words if they cannot be placed in one of the two categories (sacrificing the expressive power of the resulting subset of English). For example, if "will" the noun and "will" the verb, or "mine" the noun and "mine" the possessive pronoun fall into different categories, they may be omitted (yes, even sacrificing the whole future case is tolerable).

It wouldn't be hard if there were no constraints, would it? The division has to be natural (intuitive for a native speaker; words in the first group should be considered substantially more "significant" than words from the second one) and unambiguous: no word with the same spelling should fall into both categories, regardless of position in the sentence, part of speech, etc.

  • 2
    Do mean content words vs function words?
    – Colin Fine
    Feb 20, 2017 at 12:53
  • 1
    Unambiguous is impossible, consider the sentence: "important is spelled with an a, not an e." Feb 20, 2017 at 14:37
  • 3
    Please explain exactly what you want and why. It's extremely unfair to the answerers to reject their answers for reasons you never originally specified.
    – curiousdannii
    Feb 20, 2017 at 21:56
  • 2
    Also your last criteria of putting all homographs in the same category is surely impossible considering how many there are in English. All the proposed divisions would put "will" in both sides.
    – curiousdannii
    Feb 20, 2017 at 21:59
  • 1
    @Sergey So what are you mainly after, a subset of English, or terminology? How expressive does it have to be? Why not use something like NSM if you really want it to be unambiguous? If you're choosing the subset then any of these categorisation systems would probably work, but they'd also all be useless if you're manipulating the subset in order to make the categories fit.
    – curiousdannii
    Feb 21, 2017 at 7:11

3 Answers 3


Some additional contributions of mine:

Stressed vs unstressed:

Nouns, adjectives, lexical verbs, numerals and adverbs in English all carry primary stress, whereas articles don't. Prepositions will go into either category ('of' to the second class, 'over' to the first class).

Counterexample: The only one I can think of is that auxiliary verbs are unstressed, but you want them in the first category.

Bound vs. free:

Words in the first class may stand alone in a phrase:

(1) [NP [N Humans]] are [AP [A smelly]].
(2) [NP [PN John]] is smelly.
(3) I've got [NP [# one]].
(4) Humans [VP [V smell]].
(5) I [VP [Aux will]].
(6) He [AdvP [Adv really]] smells.

English articles, like articles in most languages, are bound forms and can't appear alone:

(7) *That's [NP [Art the]]/[NP [Art a]].

In the end you may end up needing a combination of the criteria proposed to get your classes.

  • Stressed vs unstressed: this is quite like what I was looking for, primarily because it should be intuitive to the native speaker; remember that we can omit some words if we can't classify them, that's not a problem; however, I'm a bit worried about whether all speakers will agree on the stress pattern in a given sentence. Bound vs free is a bit too complex for my purposes.
    – user15253
    Feb 20, 2017 at 14:25
  • 1
    @Sergey That's a valid point, since stress can be neutralised in everyday speech (though I think the stress patterns will be more pronounced in formal, careful speech, and less variation will be present). I don't think any speakers will stress 'the' or 'a' in any situation, except the (highly marked) THE + proper noun construction, so perhaps if the criterion is rephrased as the possibility of applying stress, I think native speakers can reach agreement. Feb 20, 2017 at 14:49
  • I find the classification "stressed vs. unstressed" not that great because they don't directly reflect inherent properties of words, but rather are a result of some different underlying properties: The reason why "the", "of", "has" (as an auxiliary), ... are usually unstressed is that it doesn't make much sense to but them in focus because they are primarily functional and don't make much conentual contribution to a sentence. Stressedness vs. unstressedness will presumably roughly correspond to contentual vs. functional meaning, and I find the latter distinction more straightforward. Feb 20, 2017 at 16:44
  • 1
    @lemontree I think it depends on your perspective, but there is evidence that stress is an inherent property of words. Consider these sentences: 1a) I saw him. 1b) I saw HIM. 2a) I saw it. 2b) *I saw IT. Stressed and unstressed pronouns perform different functions in discourse, and it happens that the stressed equivalent of 'it' is THAT, rather than IT, hence the minimal pair in (2). So it makes sense that the contrast between him and HIM in (1), too, is lexical and an inherent property of the pronouns. Feb 20, 2017 at 17:12
  • 1
    @Sergey: I can think of one way in which my bound vs. free point could be made more intuitive. All the items in the first class can be answers in sentence fragments, but articles can't (What are these? Drums. // How do you feel? Happy. // etc. BUT Which one did you pick? *The.) Feb 21, 2017 at 15:44

Yielding roughly the same as the distinction between open and closed word classes, but from a more functional perspecive rather than w.r.t. productivity (i.e. how frequently are new words form based on their class, as is the case in the open vs. closed classification) is the distinction between content words and function words:

  • content words are mostly nouns, verbs, adjectives and adverbs.
    They usually refer to predications (i.e. properties of one individual or relations between several individuals) over concrete entities in the world, such as "is a dog", "sleeps", "is red", "loves Mary" or predications over predicates (such as "slowly" which modifies some moving event, which can again be seen as a predication over entities).
    Of course, abstract nouns such as "theory", which do not really refer to a concrete entity, mass nouns such as "water" which can not discretely be counted or some other problematic cases do belong to this category too, but speakers have quite good intuitions about these words still belonging to the category "noun".
    Since content words are used to refer to something in the actual word, they are rather productive, i.e. you can always invent new words (or build them up from existing ones) to fit the language to the circumstances you want to describe, such as "bus stop" ("bus" + "stop").
  • function words comprise, roughly, categories such as articles, quantifiers, pronouns, adpositions, conjunctions and subjunctions, and are words that do not predicate something in the world but are rather there to express functional aspects such as relations between sentences and phrases ("and", "or", "if", "because", ...), quantification (such as "a", "2", "many", ...), reference ("he", "there", "then", ...), interjection ("yes", "no", "hey", ...) and others.
    Since the set of concepts such as logical relations, quantities etc. are rather limited, the set of function words corresponds more or less to the set of closed class words, as new expressions are rarely invented.

Note that, although languages may vary in the size and distribution of their lexical inventory (for example, some languages may have adjectives that behave very verbal and in which adjecties might not be seen as a separate class, or numerals which have more properties of a noun than of something determiner or adjective-like, or a rather productive and open-class set of adpositions, ...), the basic distinction between content words and function words, as well as the correponding word classes, is something more or less universal, and not specific to the English language.

  • This is coser to what I was thinking of, but again numbers fall in the second category, which is not what I need, and the categories are defined somewhat fuzzily, e.g. some adverbs act as function words. What I'm looking for is probably some sort of a "core subset" of the function words as my second category, which are so generic that they almost have no meaning by themselves.
    – user15253
    Feb 20, 2017 at 14:17
  • 1
    @SergeyThese (and the ones mentioned by other answers) are the classifications that are common in linguistics because of certain properties of words. If you want a specific interpretation into the categories that you listed and none of the classifications based on such properties perfectly suits "what you need", then just define your own terminology by saying "The class of blabubble words consists of the following categories:..." Feb 20, 2017 at 16:29
  • defining your own terminology is fine in linguistics, but I'm looking for something that should be known and more or less intuitive to English users already. As I currently figure out from these answers, there's no existing category that fits all criteria, but some are close.
    – user15253
    Feb 21, 2017 at 6:36

There is indeed some kind of classification along the lines you sketch, it is open versus closed word classes.

  • Open classes easily accept neologisms and contain a lot of different words
  • Closed classes contain a usually small number of fixed words and are resistant to neologisms

Note that the notion of open and closed word classes depends on the language you are looking at. There are language outside where the pronouns are an open class and there are languages where adjectives or verbs are a closed class.

For English, the closed classes are articles and determiners, pronouns, prepositions, conjunctions, and particles (words like "yes", "no", "not").

  • That's close, but this "determiners" class is what's problematic. It includes numerals which I need in the first class. Did you also mean "articles and determiners"? Adjectives seem rather "open"... Also, for the forseeable future English is the only language that I'm doing this for.
    – user15253
    Feb 20, 2017 at 10:22
  • Look at this answer for the meaning of determiner: linguistics.stackexchange.com/questions/12777/… Numerals aren't determiners, the Cardinals are a part of speech of their own and the ordinals are classified as adjectives (usually) Feb 20, 2017 at 10:26
  • @Sergey Adjectives are open for English, but this is not a linguisitc universal holding for all other languages of the world. Feb 20, 2017 at 10:27
  • thanks, but e.g. dictionary.cambridge.org/grammar/british-grammar/… explicitly lists e.g. "five" as a determiner; that's not the only place where I saw them included; is that an error?
    – user15253
    Feb 20, 2017 at 10:38
  • 1
    @Sergey This is a different opinion, at least. There are lots of different part of speech classifications in circulation. Feb 20, 2017 at 10:44

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.