¹ Italian requires the subjunctive for expressing an opinion, even something as positive as I am convinced that God exists.

² For example, figure, which patterns in some respects with think, reckon, fancy, etc. (He’ll do it, I think, I figure, I fancy), is, so far as I know, unique in taking how rather than why in the question How do you figure thát? = Why do you think thát?

³ Syntax (Boston, 1931), p. 248.

⁴ Semantics: Studies in the Science of Meaning (New York, 1900), pp. 230–32.

⁵ The deletion of that is difficult to formulate. With a following be it can occur more readily: It is desirable (that) he be given a second chance.

⁶ The case of believe, guess, conjecture, imagine, etc. is slightly different, as example 84 suggests. They resemble reckon etc. in not admitting or barely admitting a sentence like *That he did it I imagine, but postposition with that is not ruled out if something is done to reinforce the meaning of the verb: That he did it I can damn well imagine.

⁷ Though conducive questions are the most likely candidates for this kind of blending, it is probably not confined to them. In an example like They’re coming, it’s quite definite, both It’s quite definite (that) they’re coming and They’re coming. It’s quite definite are plausible sources, with the separate sentences perhaps a little more so. Yet comparing this example with one like They’re coming, it’s definite, we are struck by what seems to be a worse comma split in the latter. It’s quite definite tends to refer to an opinion of the speaker, hence to match other PMPs like I’m sure, whereas it’s definite suggests an external decision unrelated to anyone’s opinion. The tighter nexus in it’s quite definite thus implies at the very least a support from ordinary PMPs. A similar case is that of He’s ready, there’s no doubt about it, which can be based on either There’s no doubt about it that he’s ready or He’s ready. There’s no doubt about it. If we compare this with I’m sorry about it that you didn’t get the prize, we find, however, that even though You didn’t get the prize. I’m sorry about it is grammatical, and conceivably should pave the way to *You didn’t get the prize, I’m sorry about it, the latter nevertheless contains an unacceptable comma split. It appears to be excluded by the same rule that excludes the obvious PMP in *You didn’t get the prize, I regret. Likewise He’s ready, I’m sure of it is admitted even though the source can only be separate sentences and not *I’m sure of it that he’s ready—‘being sure of it’ rides in on the crest of the normal PMP ‘being sure’: He’s ready, I’m sure.

⁸ On the other hand, I know is likely to be accented, though it does not have to be. In He’s cóming, I know, but there’s no assurance that he’s stáying, I know can occupy a low-pitched terminally rising tail.

⁹ Compare no wonder, no kidding, sure enough, true.

¹⁰ The behavior of French craindre varies according to register: “correct” usage requires the subjunctive, but conversational usage admits the indicative, apparently on the same terms as in Spanish and Rumanian. So for Did he hurt himself?—Yes, I’m afraid he broke his leg:

—Est-ce qu’il s’est fait mal?

—Oui, je crains qu’il s’est cassé la jambe.

and even more so if the context is as overpowering as in the example Did your friend Merton accept the offer?—I think he wanted to, and I believe he considered it for a time, but I’m afraid he finally decided against it:

—Est-ce que votre ami, M. Merton, a accepté cette offre?

—Il me semble qu’il voulait l’accepter et je crois même qu’il a envisagé la chose, mais je crains que finalement il a décidé de ne pas l’accepter.

For speakers who reject the indicative, the solution is perhaps to view craindre as a lexical rather than a grammatical exception: it can be assumed to mean only ‘to view with alarm’ and never ‘to note with alarm’:

*That’s how it happened, I view with alarm (I’m afraid).

That’s how it happened, I note with alarm (I’m afraid).

¹¹ Spanish morphology is even more severely restricted. Sorprendentemente is all right as a sentence adverb; alentadoramente ‘encouragingly’, divertidamente ‘amusingly’, etc. are not. I have not investigated this point but it seems likely that prepositional phrases offer a better prospect, on the order of English much to our surprise (para sorpresa nuestra); hence para estímulo nuestro ‘encouragingly’, para desilusión nuestra ‘disappointingly’. There are other devices in English too, of course, e.g. It won’t last long, more’s the pity.

¹² Bolinger, Dwight, Interrogative Structures of American English (University of Alabama Press, 1957), p. 94.Google Scholar

¹³ For example, George Lakoff, in an unpublished pre-print, “Pronominalization and the Analysis of Adverbs” (July, 1967), p. 6, with references to Charles Fillmore and E. S. Klima.

¹⁴ The idea that in I don’t think he’s coming we have a negative element that belongs fully to the subordinate verb and that can be transferred, like a syntactic ping-pong ball, to another position without altering its logical connections, I think is not quite true. If it were true, we ought to be as satisfied with I don’t fancy (guess, reckon, calculate) he’s coming as with the same sentence using think (believe, imagine). We saw earlier that the latter set of verbs admits a that complement (Why do you think thát?) whereas the first set does not (*Why do you fancy thát?); the reason is apparently that the second set retains some of its lexical meaning (think refers to thinking) while the first set has lost it (reckon does not refer to reckoning, as in reckoning one’s accounts). This is corroborated in three other ways: (1) Reckon etc. is limited to non-progressive forms despite the fact that (unlike believe, which is stative through-and-through) its lexical source is not, whereas think etc. is not so limited: I’m just thinking that nobody’s going to turn out to fit that job; *I’m just reckoning that …; George, I’m thinking, is the man for the job; *George, I’m reckoning. … (2) As will be noted in the text, a question like Why do you reckon he’s guilty? is unambiguous—it asks only why he is guilty; but Why do you think he’s guilty? also asks why you think so and can be answered Because I’m prejudiced. Similarly Why did I write him that dunning letter, do you reckon? is a question that the speaker probably asks of himself—what were his own motives?—do you reckon only signals that the question is being pondered, not that a second party’s opinion of the first party’s motives is being sought; but Why did I write him that dunning letter, do you think? does ask the second party’s view of the first party’s motives—it is a common way of suggesting that the second party does not need to be asked, that he should have sense enough to know. (3) Reckon etc. is ungrammatical if subordinated to another verb, e.g. *Why do you prefer to reckon he’s guilty? All in all, it appears that not is received more readily by verbs retaining a sense that can be negated, and less readily by verbs such as reckon, fancy, and calculate, which have lost their lexical bearing. This can also be seen in the ungrammaticality or near-ungrammaticality of *He isn’t coming, I don’t reckon (fancy, calculate, guess) beside the grammatically of He isn’t coming, I don’t suppose (think, believe, imagine). The conclusion seems to be that the negation belongs to the main verb, hence to the PMP also, as much as to the verb in the noun clause. It does not merely hop from one to the other but belongs semantically to both.

¹⁵ With other interrogative words, lexical incompatibility would rule out ambiguity: Where do you think he put it? is hardly answerable by *I think it at home.

¹⁶ In Old Spanish the indicative was permissible after attitudinal verbs, the notion then being one of the source of the attitude rather than its projection. Thus Me alegro de que no hay más would have meant “There being no more causes me to rejoice” rather than “I’m glad that there are no more,” “I applaud there being no more.” This continues in many modern dialects, and affects the speakers of the standard to the extent that when a modal referring to a future action predicts its inevitability, the indicative is accepted. Thus speakers who may reject *Me alegro de que él llega mañana and demand the subjunctive llegue, will accept Me alegro de que él va a (ha de) llegar mañana.

¹⁷ From Poston, Lawrence, et al., Continuing Spanish, Vol II (New York, 1966), p. 6.Google Scholar

¹⁸ From Tomás, T. Navarro, Entonación Española (New York, 1944), p. 92, quoting Ramón Mesonero Romanos.Google Scholar