I am currently working on the following phenomenon:

(1) a. If (you are) hungry, you should say so.

b. He touches his nose when (he is) exaggerating.

c. Where (it is) cheap, watermelon sells well.

d. Tom hums to himself while (he is) working.

The appearance of the material in parenthesis is optional in each case. My question concerns the analysis of this phenomenon. Does anyone out know what it is called? Is it considered a form of ellipsis? Is there an established analysis of the phenomenon?

  • Note that this is close to the recently famous Word of the Year winner: Popular Prepositional Because. E.g, He hums to himself because (he is) working. But note also *He hums as far/long/soon as able to, but He hums as far/long/soon as he is able to. There's a lot of fun to be had running these through a list of conjunctions.
    – jlawler
    Commented Feb 23, 2014 at 19:48

1 Answer 1


I'll begin with a mini-rant: paraphrasing one of my former professors, whenever someone uses the term "ellipsis", I check that my wallet is still where I think it is. As far as I'm concerned (and I spend a lot of my waking hours thinking about ellipsis), "ellipsis" is a term that should be dropped from the terminology of linguistics, because it's more confusing than it is helpful. It simply means that some words are missing. But what exactly is the mechanism that makes them go missing? Even staying within the core of Chomskyan syntax, I can think of at least three different ways in which to account for "missing words". If you start considering various extensions to Chomskyan syntax, or even other frameworks, things can become more intricate.

On to your question, there are various funny things to notice:

  1. "Ellipsis" always necessarily happens in an embedded clause. Compare "When exaggerating, he touches his nose" vs. *"When he is exaggerating, touches his nose".
  2. If the remnant of "ellipsis" is a verbal predicate, it is necessarily non-finite, e.g., *"When exaggerates, he touches his nose".
  3. The implicit subject of the non-finite, embedded clause is necessarily correferential with the subject of the embedding clause. Thus "When exaggerating, he touches his nose" can't mean "When {I/you/she/we/they} exaggerate(s), he touches his nose".
  4. In fact, the previous point correlates with the attachment site of the "elliptical" clause. Consider "Adam says that Betty scratches her nose when nervous". If we take "when nervous" to mean "when Adam is nervous", then only a high attachment site is possible ---that is, "when nervous" modifies the time of saying, not the time of scratching. Conversely, if "when nervous" means "when Betty is nervous", then "when nervous" necessarily modifies the time of scratching, not the time of saying.

Off the top of my head, I'd say that you are looking at a case of non-finite predication (a "small clause") with a null controlled subject. I would call it PRO, but you can might want to call it something else, depending on your theoretical preferences. If so, a good approximation of the syntax would be:

[when [PRO_i nervous]] [he_i scratches his nose].

A proper analysis will have to be more detailed, but I bet this is a good place to start.

  • Well put, I couldn't agree more. Ellipsis comes out of the theoretical confusion introduced by Chomsky. It does not exist in syntax as such, although, it exists in discourse. Commented Feb 23, 2014 at 13:09
  • 1
    Great answer. A couple of points: I'm not sure it's quite right to call these things 'embedded clauses' - they seem to pattern more closely with sentential adjuncts. Note that this is also possible in if-clauses: "If nervous, he blinks a lot.". It also doesn't seem quite right to state that the subject of predication is always co-referential with the subject of the matrix clause. For example: "When nervous, it seems that John blinks a lot." It seems to me that this is acceptable where John is the subject of predication, despite the fact that the subject of the matrix clause is an expletive.
    – P Elliott
    Commented Feb 23, 2014 at 16:06
  • 1
    @DominikLukes: I'm afraid you are mistaken. Chomsky has nothing to do with this. There are good theory-independent arguments to the effect that certain types of ellipsis are syntactically derived (i.e., as non-pronunciation of a certain portion of syntactic structure). Some of these arguments have been developed by people (e.g., Ivan Sag) who otherwise have been very vocal opponents of the basic assumptions of Chomskyan syntax.
    – Koldito
    Commented Feb 23, 2014 at 18:03
  • 1
    Yep, very much agree. See for example Ross' 1969 paper on sluicing ("guess who" - babel.ucsc.edu/~hank/guess.who.pdf) where it seems reasonably clear that the construction analysed involves silent syntactic structure. When people talk about 'ellipsis' in generative grammar, they're generally talking about surface anaphora in the sense of Hankamer & Sag '76. In my experience most of the confusion originates outside of the field.
    – P Elliott
    Commented Feb 23, 2014 at 18:31
  • Thanks all for the answer and the comments. The important thing from my point of view is that none of you has stated that there is an established account of the phenomenon addressed in the question. This confirms my suspicion, so I'm pleased about that. Commented Feb 23, 2014 at 19:01

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.