Guest posting on “How does he sleep at night?”

Back in my 9/4 posting “How does Wilderrama sleep at night?”, I appealed to my colleagues in semantics:

Dictionaries seem not to do a lot of sense differentiation for how — NOAD boils the relevant OED2 entry down to ‘in what way or manner; by what means’, all as one sense — and I don’t know anything in the semantics literature that covers this territory (but then I’m basically pig-ignorant of the semantics literature [this is hyperbolic, meant to be entertaining, but the truth is that what I know of semantics is unsystematic and patchy, things I’ve picked up as a syntactician outsider], so I’m sketching  a treatment improvisationally here. I would be happy to be illuminated.

And got a fine thoughtful response from Hana Filip (Professor of Semantics in the Dept. of Linguistics at Heinrch-Heine-Universität Düsseldorf), in several installments, which I will incorporate (lightly edited) below, with her permission, as a guest posting.  I induced her to agree to release this material, despite its being preliminary and far from publication style, with this appeal:

The point would be not that [you and I are] hammering out an answer, but that we’re thinking things through. My readers are mostly not linguists, so they don’t get to see linguists at work, before things turn into published material. In particular, they don’t get to see us using a conceptual apparatus (and the technical vocabulary to go with it) to try to make sense of the world; we don’t inject that technicality into the discussion just for show — but much of it is really necessary to get past the point of just accumulating judgments and examples from texts.

That will involve  my contributing explanatory text at various points, to fill in things that Hana and I take for granted (on her side, she’s not a specialist in syntax, but shares a considerable body of conceptual background with me in syntax). The result is then something like a public conversation between us, with me turning to the audience every so often with commentary on the proceedings.

Background 1. The 9/4 posting presents a joke from a tv show involving two characters M and T, a joke turning on an ambiguity in interpretation, with M and T at cross purposes over their understandings of the question How do you sleep at night? (with interrogative how):

(super-conscientious) M (on learning that T has accumulated over 98,000 unread e-mails because he just couldn’t face up to coping with them): How do you sleep at night?

(laid-back, easy-going) T: On my back. Naked. Fresh air feels nice.

I then wrote:

You can appreciate something of the range of senses for interrogative how by considering four different ways of answering the question How does he sleep at night?:

truth-functional + means how in: How does he sleep at night? By taking Sominex.

truth-functional + manner how in: How does he sleep at night? Restlessly.

truth-functional + state how in: How does he sleep at night? On his back. Naked [T’s insouciant response]

modal + means how ‘by what means is it possible’ in: How does he sleep at night? By ceasing to care about his responsibilites. [M’s incredulous question]

Note 1. The modal + means how can have the modality made explicit: How can he sleep at night (after what he’s done)?

Note 2. Related to state how are two senses that are listed separately in many dictionaries, including NOAD: condition (stative) how (How do you feel? Sick.) and quality (active) how in How did they play? Very badly.)

Background 2. My discussion of sentence meanings above draws on the technical literature in semantics, but with merciless distorting simplifications for the purposes of presentation. I take it for granted that semantics as an intellectual activity concerns itself with expressions in natural languages by associating them with descriptions in some sort of special metalanguage.

By such means, different expressions, even expressions that differ in syntactic category, that convey similar meanings can be treated as instances of the same element in this metalanguage; for instance, the adjective necessary of  It is necessary that you leave immediately and the verb must of You must leave immediately share a qualification of necessity on your leaving, represented in the (now-standard) metalanguage of modal logic as ▢p, where p is the proposition that you leave and ▢ is the necessity operator.

And by such means, ambiguous expressions can be treated as instances of different elements in the metalanguage; for instance, the ambiguous must of You must eat a lot of blueberries, with the proposition that you eat blueberries qualified either as an epistemic necessity (‘It must be that you eat a lot of blueberries’ (because I see that your teeth are stained blue)) or as an obligation (‘You have to eat a lot of blueberries’ (if you want to stay healthy)) — a distinction representable in modal logic as ▢q, where q is the proposition that you eat a lot of blueberries, versus Oq, where O is the obligation operator.

In my brief analysis of the joke above, the modal operator in the meaning M intends to convey is the (epistemic) possibility operator ◇, the operator of can ‘be able to’ (versus can ‘be permitted to’, with the permission operator P): compare ◇r in I can drink milk now, because I’m taking medication for my lactose intolerance) versus Pr in I can milk drink now, because it’s just been added to my Wonder Diet. In the joke, M asks by what means ◇s, where s is the proposition that T sleeps at night.

Meanwhile, in the joke what T responded with is an assertion of the state in which unmodalized (“truth-functional”) s — what you might think of as “just plain s” — holds.

(I’ve silently folded both adverbial modification and interrogation into my treatment of the M – T exchange, though in a careful analysis, places for them too would have to be found in the semantic metalanguage.)

For the reader’s reference, I’ve prepared a cheat sheet (hat tip to Arika Okrent) of the four modal operators I referred to above, arranged in two dimensions: universal  ▢ and O (on the left) vs. existential ◇ and (on the right); and epistemic (having to do with belief and knowledge) ▢ and ◇ (at the top) vs. deontic (having to do with rules and norms) O and P (at the bottom):

(#1) The square of modality

(For historical reasons, in philosophy the plain term modal logic customarily refers to the logic of necessity and possibility, sometimes extended to include moral modal logic (of obligation and permission) as well. But there are still more modal logics, as you can see by consulting the Stanford Encyclopedia of Philosophy entry on “Modal Logic” (revision of 9/8/18).)

(A necessary footnote: in the analysis of English syntax there’s a small subcategory of auxiliary verbs traditionally known as modal auxiliaries, or simply modals. I stress that this is a syntactic category of a particular language, with its members lumped together — the core members are the lexical items can, could, will, would, shall, should, may, might, must — because of their shared syntactic properties in English. They get the traditional label modal because so many of their occurrences convey a semantic modality. But not all do; and semantic modalities are conveyed by a stunning array of lexical items in other lexical categories: for example, non-modal auxiliary verbs like have to; predicate adjectives like necessary; adverbs like obligatorily and maybe.

It ought to be unnecessary to point this out, except that if you search for information on semantic modality in English, you’ll be flooded with stuff about the English modals, material that is largely beside the point.)

Just for fun. Semantic modality is a topic singularly lacking in visual interest, so I’m appealing to the Holy Modal Rounders, who knew how to goof around in several sensory modes, including the visual. For the ear, there’s their excellent poetic name, and their music; and then for the eye:

(#2) Their first album (1964)

From Wikipedia:

The Holy Modal Rounders was an American folk music group, originally the duo of Peter Stampfel and Steve Weber, who began performing together on the Lower East Side of New York City in the early 1960s. Their unique blend of folk music revival and psychedelia gave them a cult-like following from the late 1960s into the 1970s.

That’s modal as in the modes of musical analysis, different ways of forming a scale — only distantly related to modes of meaning and to modal logics. But I couldn’t just disregard a musical group that put out an album (their fourth) entitled The Moray Eels Eat the Holy Modal Rounders.

Hana Filip on the M – T exchange. From e-mail to me on 9/8:

I looked around a bit, but did not manage to find anything published on this phenomenon. [AZ: Well, we can still hope something will turn up.]

There are examples of the modal use “M’s modal (specifically, epistemic) + means how ‘by what means is it possible’ ” that seem to be highly conventionalized, I could immediately think of examples like

(1) How could/can you do this to me?
(2) How do you live with yourself?

They allow a joking ‘Wilderrama answer’,  using one of the three truth-functional senses of ‘how’, you outline.

Quora has joking answers to (2); in them, the hearer ‘uncooperatively’ ignores the intended modal import of ‘how’, and exploits its not-intended truth-functional senses.

[AZ: And now Hana does something linguists do all the time: we cast around for other phenomena that are in some way like, or just reminiscent of, the phenomenon we’re looking at; they might have important things in common — or they might be crucially different, but realizing that would be a significant advance on things, too. But to do this, you have to a big stock of phenomena (rather abstractly characterized) to riffle through mentally; that’s what our professional experience is for.]

This seems akin to ignoring the intended and highly conventionalized  indirect speech act (request) meaning of Can you pass the salt? at the dinner table, and instead of complying with the request to pass the salt, pleasantly smiling and responding Yes, I can, but not doing anything, and so pretending that the ability meaning of can is understood. On Searle’s view, the ability meaning is the basic / literal  meaning of can and the indirect speech act (ISA) meaning of request is inferred from it, but he also uses it as an example of a conventionalized indirect speech act. So ability/ISA request is not a matter of lexical ambiguity.

[AZ: Cool idea. In any case, we’re now looking at cases of conventionalization. At some level higher than the lexical; the conventionalization attaches to expressions rather than single words. And that brings us to the question of what conventionalization means in this context. ]

The modal uses of how imply [AZ: imply might be too strong a verb] an accusation, as you also suggest. At least I cannot think of any such examples without the suggestion of an accusation, which would also point in the direction of conventionalized indirectly conveyed meaning:

How can he sleep at night (after what he’s done)?
How could/can you do this to me (after I have been so kind/generous/loyal to you)?
How do you live with yourself (after what you have done)?

The accusation can also be directed at the speaker:

How can we do this to ourselves? (Are we trying to wipe ourselves from the face of the earth?)

[HF later thought:]

…  How could/can you do this to me (after I have been so kind/generous/loyal to you)?

sounds weird to me now, but I found this gem in the pop music world:

How could you do this to me / After I gave you all of me?   [song lyrics]

[AMZ digression: popular music being what it is, there might well be such a lyric, but this looks like an amalgam of two different songs, only one of which is relevant to implicit accusatory content. The relevant one would be “How Could You Do It To Me” by Regina Belle:

How could you do it to me
How could you leave me here this way, mmm…
After all we’ve been through
You just turned and walked away

From Wikipedia:

All by Myself is the 1987 debut album by American singer–songwriter Regina Belle. Released on June 1, 1987 by Columbia Records, The album features the hit singles “Show Me the Way”, “So Many Tears”, “How Could You Do It to Me”, “Please Be Mine” and “You Got the Love”.

Meanwhile, giving all of me, no accusation, is from John Legend’s love song “All of Me”: ... I give you all of me / And you give me all of you. From Wikipedia:

“All of Me” is a hit song by American singer John Legend from his fourth studio album Love in the Future (2013). It is dedicated to Legend’s wife Chrissy Teigen.]

[Hana on 9/12:]

A friend today reminded me of the angry song that John Lennon wrote as a reaction to Paul McCartney’s successful suit to dissolve the Beatles. The song is entitled “How do you sleep?” [with the refrain How do you sleep? / How do you sleep at night?] and it is clearly meant as an accusation (modal+means ‘how’). See the Wikipedia article.

For the interpretation of the ‘how’ question, it also seems to matter whether the AUX is “do” or “can/could”. How do you sleep (at night)? seems open to either one of the truth-conditional meanings or the modal+means use of ‘how’. How can you sleep (at night)? — for me, there seems to be a preference for the modal+means how interpretation. The truth-conditional meaning is also possible in the right context; for example, a friend complains about problems at work, being in debt, and a number of other calamities … then I’d use How can you sleep (at night)? to show a sincere concern for my friend’s well-being.

[Finally, Hana turns to another linguist’s strategy in trying to figure out what’s going on with some data: ask how things work in other languages. Now, we all understand that different languages have different structures, so you can’t just assume that if the analysis is much clearer in language L, then you can adopt that analysis for the language you’re looking at. But it’s worth playing with the idea — if only to clarify how the language you’re looking at differs crucially from language L.]

Btw, Italian has two different constructions, (i) one that has a distinct preference for the modal+means how and (ii) the other that seems to admit only one of the appropriate truth-functional ones:


[AZ 9/21: We are both still at the musing stage.]


Leave a Reply

%d bloggers like this: