"… to do is (to) VERB …"

« previous post | next post »

Dyami Hayes writes to point out that there has been a change over the past century in the relative popularity (at least in printed text) of constructions like these:

What this book sets out to do is to provide some tools, ideas and suggestions for tackling non-verbal reasoning questions.

What it attempts to do is provide a framework for understanding how local governments are organized.

The Google Books ngram plots for provide, look, tell, and say show similar patterns — or summed for those four verbs (with the to do is VERB version in red and the to do is to VERB version in blue):

In each case, we see a striking rise over the past century of the frequency of the to do is VERB version, starting from nearly nothing around 1900, and with an acceleration over the last few decades of the 20th century. The to do is to VERB version declines slightly in popularity, but not at all in proportion of the rise of the alternative.

But analogous data from COHA shows a different relationship:

Here the frequency of the to do is to VERB pattern declines earlier and more steeply, while the to do is VERB pattern rises earlier and more gradually.

In either case, there's been a change in a well-defined direction. Why, and why the apparent different is dynamics, is a topic for another day.



  1. Ken Miner said,

    December 10, 2015 @ 9:39 am

    Factor in the double-copula phenomenon and your Ngrams might look different: to do is is to VERB

    [(myl) Not really: "to do is is to [v*]" doesn't occur at all in COHA; and strings like "to do is is to tell" etc. have length longer than 5, so the Google Ngram viewer will not show them. But in fact "to do is is to" is not found in the Google corpus at all (presumably because there were never more than the threshold of 40 in any year) So in sum, ISIS is not a factor.]

  2. Neil Dolinger said,

    December 10, 2015 @ 12:02 pm

    I suppose that an alternate construction that many style guides would favor would be,
    "This book sets out to provide some tools, ideas and suggestions for tackling non-verbal reasoning questions. "

  3. Mara K said,

    December 10, 2015 @ 12:55 pm

    Reminds me of this quote, which I first saw on a T-shirt in high school.

  4. Jerry Friedman said,

    December 10, 2015 @ 2:59 pm

    Out of curiosity, I repeated the search with "should do is…" My hypothesis was that the crossing point would be earlier, since the first "to do" would prime the second. It was disconfirmed.

    My new hypothesis is that there's some effect like dissimilation that caused a few people to avoid repeating "to".

    I wonder how turning these around works. I can kind of see

    ?Tell them at once is what we should do

    but not

    *Provide some tools, ideas and suggestions for tackling non-verbal reasoning questions is what it attempts to do.

    ("Ease my troubles, that's what you do" is a little different.)

  5. Jerry Friedman said,

    December 10, 2015 @ 3:00 pm

    My attempt to embed the chart didn't work. Here's the link:


  6. CD said,

    December 10, 2015 @ 3:32 pm

    Well from a purely stylistic standpoint I would suggest just

    "This book provides tools, ideas and suggestions for tackling non-verbal reasoning questions. "

    which is surely a true statement i.e. if it set out to do that and failed, one hopes it would not be published at all. Moreover "provides tools, ideas and suggestions" is still politely tentative, if that's the effect you're aiming for. The extra dos and tos are just throat-clearing.

  7. BZ said,

    December 10, 2015 @ 5:33 pm

    I don't have any explanations, but repeating the to sounds completely wrong to me.

  8. Guy said,

    December 10, 2015 @ 9:19 pm

    One thing I've always found interesting about this construction is that the complement of the copula (with or without "to") can't syntactically fit in the gap unless we say this is some specially licensed use of do that can only appear in relative clauses. I wonder if this change is reinforced by the fact that if this were supportive do (which seems hard to argue, but still) then the to-less version actually would syntactically fit.

  9. Jerry Friedman said,

    December 11, 2015 @ 12:57 am

    Actually, you get a very similar ngram result without anything before the "do". The hits at GB seem to mostly have auxiliaries before the "do": "the best we can do is [to] VERB" and the like.

    After "what they do", the form with "to" is not found, as in this chart. The ones without the "to" are fairly recent, starting in the early '60s—coincidence?

    BZ: I think I'd always say it without the "to".

  10. Jerry Friedman said,

    December 11, 2015 @ 1:13 am

    I think a good thing to do is search the GB American corpus using the BYU interface (I didn't know you could do that!), but it's late and I don't have those nice graphing tools. A search for [p*] [do] is to [v?i*] seems to return lots of relevant stuff, mostly like "all you do is (to) VERB".

  11. Tim Friese said,

    December 13, 2015 @ 12:06 pm

    A few more notes: I think we can add 'help' to this list where there has been change from 'help to V' to 'help V'.

    Secondly I think there is a Br v. Am difference here. Frequently when I read what I find to be a clunky usage of 'to', I check the publication/author I'm reading and it's British.

  12. Guy said,

    December 14, 2015 @ 5:05 am

    @Tim Friese

    With "help" there is a subtle semantic difference. "I helped him teach the class" suggests that I was directly involved in the teaching, that I did some teaching myself, whereas "I helped him to teach the class" can be reconciled with indirect assistance (maybe I gave him some pointers on teaching techniques) more easily than the former sentence.

RSS feed for comments on this post