November 26, 2007

More on double modals: the Case of the Missing Infinitivals

In what I admitted was merely an idle speculation, I recently suggested that before we conclude that locutions like might could are examples of modal auxiliary verbs followed by modal auxiliary verbs, we need to rule out the simple possibility that might has turned into an adverb for some speakers, just as maybe did a long time ago. Some people have written to me to say this seems exactly right. Others have pointed out major difficulties for it. It would probably be appropriate for me to say a little more about the issues that bear on this. I will try to do a post or two on the topic. This will be the first. Read on if you wish to hear what I have. This first episode is about the strange case of the Missing Infinitivals.

Somebody wrote to me to say that Dutch allows double modals all over the place: dat zou niet moeten kunnen", literally translates as "that should not must can-be", and more liberally translates as "that ought not to be possible"; of je dat zou moeten willen doen translates literally as "whether you that should must want-to do", i.e., "whether you should desire to do so". This suggests that I have not made the point about double modals clear enough.

Dutch has verbs that are cognate with the English modal auxiliaries, and so does German; but there is a major difference. The English ones only have tensed forms. That is not true of the Dutch or German verbs, so they are entirely irrelevant. The most important of the special features that characterize the modals (also known as the special finites or anomalous finites) is illustrated by the contrast between the regular transitive verb can of can tomatoes and the modal auxiliary can of can go to Spain. In what follows, an asterisk in front of a string of words means it is not a grammatical sentence.

PRESENT TENSE:   Often we can tomatoes.   Often we can go to Spain.
PRETERITE TENSE:   Back then we canned tomatoes.   Back then we could go to Spain.
PLAIN FORM:   We hope to can tomatoes next year.   *We hope to can go to Spain next year.
PAST PARTICIPLE:   For a long time we have canned tomatoes each summer.  *For a long time we have canned go to Spain each summer.
GERUND-PARTICIPLE:   We are canning tomatoes right now.   *We are canning go to Spain right now.

I hope the point is clear: when the syntactic context permits tensed forms, the modal is fine; but when the context demands an untensed form — in infinitival clauses, where the plain form is needed; in a perfect tense clause, where the verb after have must be past-participial; or in the progressive aspect, where the verb after be must be gerund-participial — the modal is banned. It simply doesn't have a plain form or a past participle or a gerund participle. All the modals are like that.

Now, the thing is, in Standard English the verb following a modal is required to be in the plain form. So it immediately follows that in Standard English a modal can never follow a modal. We get this pattern:

FOLLOWING MAY:   They say we may can tomatoes.   *They say we may can go to Spain.
FOLLOWING WILL:   I think we will can tomatoes.   *I think we will can go to Spain.

So this is one point to make about double modals in the dialects that supposedly have them: if we find modals following modals, we should also find that modals can occur in infinitival clauses. That is, we should find not only I might can fix it but also (and I don't know whether to prefix these with an asterisk for the relevant dialects or not) [*?]I hope to can fix it, and [*?]I want to can fix it, and [*?]To can fix it would be nice, and so on. But although people report having heard might can, they never seem to report hearing hope to can. One of the first things I would want to know about the alleged "double modal" dialects would be, why not? Where are all the cases of modals in those other plain-form contexts, like infinitival clauses?

[Update: I'm not claiming sequences like "hope to can" are never found anywhere, of course. Michael Wescoat tracked down a couple of potentially genuine cases:

As an educator I hope to can encourage African-American students to consider nursing as their choice for higher education and future career and profession.
[http://www.coe.ohio-state.edu/beverlygordon/gordon/courses/863/nurses.html]

I hope to can keep a running journal of my adventures to keep me organized, and for your enjoyment!
[http://www.ringsurf.com/netring?action=info&ring=MDCknitring]

But as he says, we cannot tell whether these were just typing or editing errors. My point is that a serious piece of work on double-modal dialects would have to determine whether modals (or at least some of them) freely occurred in infinitival contexts. Either the syntax of modals is quite different, and modals don't take plain-form verbs in those dialects (which seems strange, for surely everyone accepts It may be raining rather than *It may is raining), or they do take plain-form complements, in which case we should find modals in all plain-form contexts. To study double modals, you can't just concentrate on collecting double-modal examples and pinning them to a board in a display case; there's a whole syntactic and morphological ecology you have to study.]

Posted by Geoffrey K. Pullum at November 26, 2007 10:55 AM