I have seen some debate on EL&U about whether or not sentence fragments are acceptable to use, and under what circumstances. I am not of the persuasion that they should be used, but if I must concede that they are permissible to use, I would like to know the rules that render them acceptable. For instance, it seems that most "acceptable" sentence fragment constructions function in a similar way to pronouns: they refer back to an antecedent. With pronouns, the antecedent is a noun, and with sentence fragments, the antecedent is a subject-verb group. Here is an example to clarify:

"The dog bit the child five times. Every day."

While I would prefer to solve the issue by using a dash ("The dog bit the child five times – every day"), how would one who allows sentence fragments justify the construction? What are general rules by which one can determine the acceptable construction of sentence fragments? My intuition tells me that not even fragment-lovers would accept constructions such as "The dog bit the. Child five. Times every day." . . . on what grounds would such constructions not be acceptable? To state it positively, on what grounds does one determine the grammatical acceptability of a sentence fragment group? By "sentence fragment group" I am revealing a presupposition that a sentence fragment derives necessary contextual data from a nearby complete sentence.

  • 1
    Perhaps thinking of sentence fragments as adopting the antecedent noun-verb group from a prior sentence is insufficient; re-working the example sentence into "The dog bit the child five times. The dog bit the child every day." seems to misrepresent the original intention by implying, "The child has been with the dog for five days, and each of those days the dog bit the child once." The original example, though, implies, "The dog bit the child five times per day, for however many days the dog has been with the child."
    – Medialific
    Commented Oct 8, 2013 at 1:40
  • 2
    Context, sure, but not always from a nearby complete sentence. Imagine saying "two hot dogs" to a hot dog vendor, who replies "two dollars". You trade dollars for dogs, but neither of you says another word.
    – user28567
    Commented Oct 8, 2013 at 1:54
  • 2
    The rule governing when a particular practice is acceptable in English is always "When the listener is willing to accept it." I'd be surprised to hear someone has managed to identify a rule of thumb, as all the acceptable sentence fragments I've encountered have been context-dependent, idiomatic, or both.
    – user867
    Commented Oct 8, 2013 at 3:22
  • 1
    Don't forget, rules for grammaticality have to do with Spoken language. That's what native speakers learn long before they learn to read and write, if they do. And in spoken language, there is no punctuation and no capital letters and you don't know where one official sentence begins and the last one ends. If it does. Because we have afterthoughts. Which are put at the end. Unsurprisingly, as it turns out. Commented Oct 8, 2013 at 4:08
  • 1
    Just asking. The phenomenon exists in many languages. Why not discuss this on linguisticsSE? That fragment has no antecedent, only a forward reference, incidentally.
    – Kris
    Commented Oct 8, 2013 at 11:04

1 Answer 1


Word strings are fragmentized in two distinct fashions I can think of: 1) by mirroring the speaker's arbitrary stressing of words, in which case absolutely any word, even a particle, is liable to receive the stress; and 2) by mirroring the intrinsic structure of the string, i.e., the valency of its parts.

1) There are absolutely no rules in this case. Licencia poetica reigns unhampered. Even though there indeed exist the content words (lexically rich units) and the function words (the connecting units), every last word carries at least an iota of meaning, and in some situations that iota just might be important enough for the speaker to stress it. One example:

Yes, but with one small difference. The Navy launched missiles over their capital. Our major general flew. To. Their capital.

2) The string can be fragmentized only at those loci at which the stress is intrinsically conveyed owing to valency being low. The lower the valency, the stronger the stress. The semantic one, not necessarily the articulative. Therefore, one can separate adjuncts, which are always optional, but not optional arguments (well, perhaps), and definitely not complements, which are never optional. Examples:

an adjunct: They helped us magnanimously. -> They helped us. Magnanimously.

an optional argument: They helped us. —> ?They helped. Us.

a complement. We need some help. —> *We need. Some help.

(Sources: none really. It's noth but an educated guess, but I urge you to consider it.)

  • 1
    I'd say that there's often no question that a verb that can behave either transitively or transitively can be followed by a legitimate or quirky DO: And then they danced. A tangle. Commented Jun 5, 2019 at 18:52

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.