I was speaking to an English learner and said, “All you have to do is read a lot.” And they thought that sentence wasn’t grammatically correct because I dropped the word to between is and read.
They thought it should be “All you have to do is to read a lot.” That sounds weird to me.
How do I explain to them the reason you drop the to? Or am I incorrect and it really should have a to?
The reality of the language is such that both forms are used, on both sides of the Atlantic, but the bare-infinitive form is clearly preferred, as the stats from the Corpus of Contemporary American English (COCA) and the British National Corpus (BNC) illustrate:
COCA BNC
all you have to do is [inf] 842 72
all you have to do is to [inf] 17 20
The preference does not change if all is replaced with what, if an adverb is introduced before the infinitive, or if a different pronoun is used instead of you. However, what also does not change is that the variant with to at least exists. (A peculiar exception seem to be she and it; negation is another interesting case, but the sample size is sadly too small for those). Sometimes the ratio is a mere 1:60, but other times it's not anywhere as cut and dry. Here are all the stats I have compiled so far:
COCA BNC
all you/we/they/I have to do is [inf] 842/206/68/62 72/27/9/8
all you/we/they/I have to do is to [inf] 17/ 3/ 3/ 5 20/11/4/2
all he/she/it has to do is [inf] 105/40/11 5/3/1
all he/she/it has to do is to [inf] 6/ 0/ 0 2/1/0
all you/we have to do is [adv] [inf] 9/5
all you/we have to do is to [adv] [inf] 0/0
all you/we/they/I have to do is not [inf] 1/1/1/1
all you/we/they/I have to do is not to [inf] 0/0/0/0
what you have to do is [inf] 59 8
what you have to do is to [inf] 11 4
what you have to do is [adv] [inf] 8 1
what you have to do is to [adv] [inf] 1 0
So if you want to be on the safe side, bare infinitive certainly is the way to go. It also happens to be the more logical choice, as demonstrated by FumbleFingers in his answer. But we can't label the other option ungrammatical, and its existence can be explained logically as well, as metanalysis.
Try replacing have to with must...
2: What you must do is read a lot.
Not only does the first to disappear; the possibility of including a second one vanishes too.
I think it's easier if we assume these sentences are "cut down" versions of the [hypothetical]...
1a: All you have to do is you have to read a lot.
2a: What you must do is you must read a lot.
Then we see that OP is deleting just one half of the phrasal verb have to, which you shouldn't really do (it should be all or nothing when deleting "the verb"). But because people don't always consciously recognise have to as a "syntactic unit", they do sometimes split it as in OP's example.
It's worth pointing out that native speakers usually delete the repeated subject "you" as well, but it certainly doesn't sound seriously weird to me to leave it in (All you have to do is you read a lot.).
Many possible ways of looking at this present themselves to me. Let me give you a few of them:
Consider this: "Have to" equates to the now slightly archaic usage of "need." If you were to say, "All you need do is read," it would sound just fine.
Consider this: "Have to do" can also be broken up as "have" and "to do," which then means you "possess" the task of "reading," which is not functionally an infinitive. The infinitive of a verb is intended as the inclusive version of the verb; it is the blanket term for all possible forms and uses of the verb. Your task, in OP, is not somehow the "all-inclusive" version of "reading," but merely your one specific task. This makes the infinitive "to read" incorrect.
Consider this: "To do" is in a sense being "explained" by "read," so "to" is not needed because the explanation requires only the substantive part of the action, namely "read."
Consider this: Repeating the "to" before "read" would simply be redundant, hence unnecessary.
“All you have to do is read a lot.” as opposed to “All you have to do is to read a lot.”
(Much of the following references OED and Jespersen's "A Modern English Grammar on Historical Principles.")
The "to" (and its loss) is a vestige of Old English that had grammatical cases in which nouns had suffixes indicating both the dative and the form of the infinitive.
History:—As well as the simple infinitive, or verbal substantive ending in -an (Middle English -en, -e), Old English had a dative form of the same infinitive or a closely-related noun ending in -anne, -enne. In Middle English this reduced successively to -ene, -en, and then, -e, and thus eventually became the same as the simple infinitive and reduced to the uninflected verb-stem.
As an invented example: Help (verb) would have been helpan.
In Middle English, this reduced to “help”.
A/the help (noun) would have had a dative form helpanne/ helpenne
In Middle English the dative noun became -> helpene -> helpen-> helpe->help.
The infinitive and the noun were indistinguishable.
This dative form, which implied to or for {noun}, was governed by the preposition tó/to. Originally, to before the dative infinitive (e.g. to helpan (v.)) had the same meaning and use as before ordinary substantives (e.g. to the helpan (n.)), i.e. it expressed motion, direction, inclination, purpose, etc., toward the act or condition expressed by the infinitive; as in ‘he came to help (i.e. to the help of) his friends’, ‘he went to [the] stay there’, ‘he prepared to [the] depart (i.e. for departure)’, ‘it tends to [the] melt’, ‘he proceeded to speak [the speech]’, ‘looking to receive [the reception[of]] something’.
But in process of time this obvious sense of the preposition 'to' became weakened and generalized, so that tó finally became the ordinary link expressing any prepositional relation in which an infinitive stands to a preceding verb, adjective, or substantive.
The result of the above is that the "to" of the infinitive is now seen as part of the verb, and more a "particle" than a preposition.
