On Monday Nancy Friedman offered this awkward example to me:
“One of the things that you always want to be for, whether you’re a Democrat or a Republican, is that you want everyone who’s eligible to vote to vote.” – Steve Schmidt, McCain strategist in 2008 (link)
There’s nothing syntactically wrong with this sentence, but the repetition of to vote might give you a moment’s pause. Nevertheless, the first to vote is an ordinary infinitival complement of be eligible (They are eligible to vote), in a relative clause modifying everyone (Everyone who’s eligible to vote is coming), and the second to vote is an ordinary infinitival complement of want in combination with a direct object of that verb (We want everyone to vote), and to vote to vote is merely part of what you get when you put these pieces together in ordinary ways.
Repetitions like this one — repetitions I’ll call Toto examples (Totos for short), after to vote to vote and in recognition of the passing feeling of oddness that they can produce (“Toto, I’ve a feeling we’re not in Kansas anymore”) — are quite common, though a careful stylist might want to avoid some of them as distracting. But then there are syntactic constructions that specifically call for repetitions of constituents. And still other configurations that you’d expect to be acceptable — more Toto examples — that are nevertheless just ungrammatical (for some speakers).
More Totos. The Schmidt example has to + VP functioning as a V complement, in two different constructions. There are similar examples with PP = P + NP in two different constructions, in things like I want you to shout for me for me, with shout for ‘call, ask for’ in combination with the VP adverbial for me ‘on my behalf’, or It’s time to send the packages to France to France, with packages to France (having to France ‘intended for France’ as a postnominal modifier) as a direct object in combination with the indirect object to France.
Totos with single words repeated are very common indeed:
But I do do the dishes, every night. (emphatic supportive do plus main verb do)
Before I gave up, I had had many problems with them. (perfective have in PST plus main verb have in PSP)
I know that that idea won’t work. (complementizer that plus demonstrative determiner that)
Everyone who went went crazy. (first went as the verb in the relative clause, second went as the verb in the main clause)
What the problem is is your manners. (first is as the verb in the subject clause, second is as the verb in main clause) Compare Andy Griffith’s famous “What it was was football”.
Stipulated repetition. That last example, with is is (and was was), is not an instance of what’s often called “double is“; it does have the sequence is is in it, but both occurrences of is there are structurally justified as parts of other constructions. In the classic types of the non-standard but widespread construction Isis (as I’ve come to call it), there’s an introductory portion ending in a form of be, with a continuation beginning with a form of be, and only one of these forms seems to have an independent justification; instead, they must be stipulated as parts of the construction. There are two main subtypes, one in which the introductory portion has a “thingy noun” (thing, problem, idea, etc.) as its head, another in which the introductory portion incorporates the subject of a pseudocleft:
[thingy-N type] The funny thing is is your girlfriend Lisa had tanning lotion on her body too.
[Pseudocleft type] What’s nice is is that it has a sort of other-worldly character about it.
(Before you rush to comment on the phenomenon, please look at this summary material on Isis and related constructions.)
A morphological construction of English in which repetition is stipulated has come to be known as Contrastive Focus Reduplication (see especially a 2004 paper by Ghomeshi et al., discussed by Mark Liberman on LLog here):
I’ll make the tuna salad, and you make the SALAD–salad.
(distinguishing prototypical cases of salads from less central cases). The construction can take in some multi-word constituents:
Do you LIKE-HIM-like him?
Oh, we’re not LIVING-TOGETHER–living-together
Elsewhere, in syntax material is stipulated as being duplicated in many constructions, for instance the tag questions of English:
You are leaving, aren’t you?
And reduplication of many different types is richly illustrated in the derivational and inflectional morphology of languages all around the world, expressing a variety of grammatical categories.
(These phenomena don’t always result in adjacent instances of some element, but they do involve stipulated repetition.)
Anti-repetition. Though repetition serves many purposes in morphology and syntax, carrying semantic or pragmatic content — not to mention in discourse organization, where repetition is employed for rhetorical effect (see, for example, Tannen’s “Repetition in conversation: Toward a poetics of talk”, in Language, 1987) — it also presents something of a challenge in language production and perception, so a wide range of strategies are used to avoid certain cases of repetition.
In morphology, sometimes an otherwise motivated sequence XX is simply ungrammatical, and its semantics must be expressed by a paraphrase; sometimes XX is haplologized to X; and sometimes one or the other of the occurrences of X (or the entire XX sequence) is replaced by a special, suppletive, element. (Extended discussion in Menn & McWhinney, “The repeated morph constraint” in Language 1984.)
There are somewhat similar phenomena in syntax, though there’s a question as to whether the offending sequences are picked out phonologically or more abstractly, and a question about whether we’re looking at a firm constraint against certain XX sequences or merely a stylistic dispreference for them.
Both issues come up in the now-considerable literature on the so-called “Doubl-ing“ constraint of English; what I have to say here is extracted from a much longer presentation in Pullum & Zwicky, “Gerund participles and head-complement inflection conditions” (1999; available on my website here). The crucial data concern contrasts between superficially similar structures like:
Terry was enjoying reading aloud.
*Terry was starting reading aloud.
Many V-ing V-ing sequences are, at worst, stylistically awkward (they’re Totos); P&Z give a wide range of cases, including the enjoing reading example above, as well as:
We could hear the screaming coming out of the air vents.
Waldo keeps molesting sleeping gorillas.
and conclude (p. 255-6) that
The Doubl-ing constraint is in fact a syntactic condition, not a morphological or phonological one. Its presence in the grammar may perhaps owe something to a phonesthetic dispreference for jingling sequences of similar-sounding endings (see Bolinger 1979 ["The jingle theory of Double -ing"] for a claim that nothing more than this … is involved in the Doubl-ing constraint), but it has clearly been grammaticized in a very specific way …
Details in the article. In summary: there is a sort of repetition constraint here, but it applies only to a small set of situations, and these are picked out by abstract features — a Vprp in construction with a VPprp complement to it — rather than phonologically.
Overall summary: There are lots of Totos, and also a fair number of cases of stipulated repetition (repetition can do a great deal of useful work), plus some language-particular constraints on repetition, especially in morphology (repetition can be hard to produce and understand).