I've been struggling, for a couple of months now to find the term for a concept from computational linguistics. It means something like: the minimum number of words that need to be placed after a given word to complete the meaning of a phrase.

For example:

  • Apple: 0
  • See: 2
  • Dream: 3

Essentially, the phrase "green apple" is logically fully complete, so there is no need to add new words after "apple". That's why its number is 0. The phrase "I dream" is incomplete, because after "dream" we need to place an object. In "I dream about an apple", we need 3 words to complete the meaning of the phrase ("about", "an" and "apple"). In "I see an apple" we need 2 words after "see".

As I understand, this term is very close to Link grammar, from Syntax Analysis. What is the name given to this concept in the literature?

  • Are you saying that to complete the meaning of the phrase "I see", we need at least 2 words? What about "I see apples"? Commented Dec 7, 2011 at 11:28
  • You correct, it should be at least 2 words after. It's for example, in other case it can be 1 word. I'm guided by subject–verb–object(SVO) topology.
    – sigrlami
    Commented Dec 7, 2011 at 11:33
  • And of course in discourse after somebody explains something to you it is extremely common to say "I see." Commented Dec 8, 2011 at 14:15
  • 1
    I think it would be better to ask about 'obligatory constituents' rather than 'words'. Commented Aug 24, 2012 at 23:53
  • @hippietrail but after "I see" is an implied end of the utterance, along the lines of "I see what you're saying."
    – Andy
    Commented Aug 26, 2012 at 14:49

3 Answers 3


I believe the term you are searching for is the valency or transitivity of a verb. Note that it defines rather how many Phrases or Consituents a Verb takes at minimum than how many words: For example "I see you." is as correct as "I see an apple".

  • Prepositions can also be considered to have valency. Commented Aug 24, 2012 at 23:55
  • i did not know this, ty.
    – taylor
    Commented Aug 27, 2012 at 1:22
  • 1
    But looking at valency does not give you order (these two are independent), and valency only gives you the kind (and number) of arguments, not how many words are used to realize them: "an apple" in your example has two words, but the valency requirement is just one object; "it" would be as good in the right context. I think your initial description ("the minimum number of words that need to be placed after a given word to complete the meaning of a phrase") cannot be given a sound definition. Commented Jan 16, 2014 at 15:52

Though notice that valency is not a constant associated with the verb alone; it's a property of the constuction it's in. There's a vast amount of variation of affordances among verbs, and even transitivity is hard to predict. Consider, for instance, the following uses of the prototype intransitive verb walk

  • He walked Monday.
  • He walked a kilometer.
  • He walked the dog.
  • He walked her to the door.
  • He walked it off.
  • He walked himself to death.

Also, there's the opportunity of a fencepost error, because only impersonal verbs like rain have valence 0; Su is 1, Su + DO is 2, Su + DO + IO is 3.


I've heard of something, which is not the answer your looking for but nonetheless related and which I thought might be interesting, called the cohort model from my first year linguistics text (O'Grady's Contemporary Linguistic Analysis 6th ed):

This model states that, in comprehension, words are analyzed by hearers from beginning to end. So for example, when we hear the word glass we initially consider all the words that begin with with the sound [ɡ]. When the next sound [l] is recognized, the number of possible words (the cohort) is reduced to those words that begin with [ɡl]. This process continues until the cohort of possible words is reduced to one - the word being recognized.

I thought I saw somewhere else in the book the same "cohort" notion being used for something like the set of possible syntactic structures given a partially complete phrase, but I guess I was mistaken in that.

I only mean to say that "a smallest set of words that need to be placed after a given word to complete the meaning of a phrase" can be thought of as an element of something like a cohort. If you think of a cohort as the set of all words and their syntactic configurations (in a conceptually similar way to the use above) that could be placed after a given word to complete the meaning of a phrase, then "the minimal number of words that need to be placed after a given word" would be found by making equivalence classes on the cohort by equating n-tuples of words having similar syntactic structure and then noting the smallest n. So, for example, some equivalence class has the elements {(he, car, drive), (she, book, buy), (they, baby, raise), ...} with element length 3 and which is smaller than the equivalence class {(he, car, drive, they), (she, book, buy, they), (they, baby, raise, they), ...} with element length 4.

  • Well I am very impressed with your knowing Mr.O'Grady personally. You are right though, I made a mistake. I only meant to point out something which was similar that I thought interesting. I apologize, I'll change my answer.
    – taylor
    Commented Aug 27, 2012 at 0:34
  • It is no doubt true that for some people, for some words or phrases, this is correct. However, it seems equally true that this is not correct for other combinations of people and phrases, and that there is thus no reason to expect human cohort structures to be anything but idiosyncratic. There are a lot of possible combinations, enough to reflect any individual's personal language history.
    – jlawler
    Commented Oct 27, 2012 at 2:44

Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.