X is the same as Y

It’s been several months since I’ve posted anything, mostly because I’ve been busy with coursework, research, TAing, etc. But I have been keeping some squib notes. Here is one.

In formal methods class one day, my prof. wrote on the board two sets:

X = \{a,b,c\}
Y = \{1,2,3,4\}

After a few seconds, he realized something was wrong, and said, “Actually, X is the same as Y.”

My initial reaction was to think, “Ah, then X=\{1,2,3,4\},” or maybe “OK, then X=\{a,b,c,d\}, where a=1, b=2, c=3, d=4.” In other words, add an element to X to be like Y.

Turns out that what the prof. meant was that Y=\{1,2,3\}, where a=1, b=2, c=3. In other words, subtract an element from Y to be like X.

It made me wonder whether, given any generic sentence like “X is the same as Y”, one could tell whether (one’s conception of) X ought to change to conform to (one’s conception of) Y, or vice versa. For me, I think X is the one to change, not Y.

For example, let’s say you think that John is 5’6” tall, and Sue is 5’8”. If I tell you, “Actually, John is the same height as Sue”, can you tell how tall they are? If I were to actually utter that sentence, I think it would be to convey that John is 5’8”, i.e., (your conception of) Sue’s height. Likewise, to convey that Sue is 5’6”, I’d say, “Actually, Sue is the same height as John.” In other words, the subject of the sentence is the one whose height needs adjusting.

Same goes for lots of other scenarios: “John is as old as Sue” to me means, basically, “However old (you think) Sue is, that’s how old John is, too” and not “However old (you think) John is, that’s how old Sue is.” It holds for comparatives, as well: “John is older than Sue” means “However old (you think) Sue is, John is older than that” (i.e., adjust/raise John’s age to exceed Sue’s; don’t adjust/lower Sue’s age to be less than John’s).

The generalization, I think, is that if I want to correct someone’s conception of X to conform to Y, I say, “X is (the same as/as [adj] as/more [adj] than/…) Y”, and not “Y is (the same as/as [adj] as/more [adj] than/…) X”.

Of course, it’s only a generalization—perhaps a pragmatic, conventional one, which can be flouted. If I’m a landscaper and you hire me to make your front yard as pretty as or prettier than your neighbor’s, and I go and dig holes in your neighbor’s yard and dump trash all over it, you’ll probably say, “That’s not exactly what I had in mind”, but I could probably argue that I successfully completed the task.

Posted in Uncategorized | Leave a comment

is is that

TED recently posted a talk by Kevin Slavin entitled “How algorithms shape our world”. The subject matter is interesting, but even more interesting, for me at least, is a pervasive peculiarity in Kevin’s speech: in constructions like “the problem is that” and “what could go wrong is that”, he often (though not always) repeats the verb is, saying instead, “the problem [with that] is is that” (10:08) and “what could go wrong is is that” (4:59).

Other examples from the talk are:

the magic and the horror of that is is that (4:15)

and that’s the thing, is is that (5:31)

and what you see here, or what you don’t really see, normally, is is that (8:54)

and what this map says is is that (13:34)

Curiously, the repetition is missing from other, similar examples, two of which are given below.

the gag is that (6:31)

the thing you might’ve noticed is that (13:48)

Note also that, so far as I can tell, when the optional complementizer that is absent, or in cleft constructions followed by noun phrases or question clauses, is is never repeated.

what he’s done is he’s actually reshaped (0:28)

what you see, that precipice […], is the 2008 financial crisis (0:38)

what you can picture is a bunch of algorithms (4:34)

the question is, […], what was happening (7:04)

what you see is the evidence (7:13)

My initial hypothesis for something like this would normally have been prosodic: there’s a prosodic break between the two is‘s. In fact, I think this is quite common. For example:

What could go wrong is, [pause], is that …

The [pause] could perhaps also be substituted by something like then or of course.

But that’s not what we have here. There is no pause between Kevin’s is‘s; they’re spoken in quick succession. Plus, there’s at least one example where a pause would, I think, be totally impossible:

that’s the thing, is is that (5:31)

*that’s the thing is, [pause], is that

What’s more, Kevin actually does not have the repetition in precisely the place where the prosodic explanation would be likely to put one: he puts a clear pause between is and that in “the gag is that” (6:31).

So it’s as if the repetitive is is functions as a unit. It’d be one thing (peculiar yet understandable) if he did it all the time, but it’s curious that there seems to be some free variation between is that and is is that.

[NB: There is one instance where Kevin seems to have not two, but three(!) is‘s:

and the thing is is is that (8:46)

But maybe I’m just mishearing him. Maybe he’s actually saying something like “and the thing of(?) this(?) is that”, despite the fact that “the thing of this”, instead of, say, “about this”, sounds unnatural to me.

Some evidence that “is is is” may be “of this is” comes from an earlier part of the talk, in which he, I’m pretty sure, says:

the point of this is that (1:02)

Here, “of this” is perfectly natural.]

Posted in Uncategorized | Leave a comment

re: Distant Drums (a reply to Geoff Pullum)

Geoff Pullum over at Language Log wrote a post about the Jim Reeves country song “Distant Drums”. His question was about the underlined part below.

So Mary marry me, let’s not wait
Let’s share all the time we can before it’s too late
Love me now, for now is all the time there may be
If you love me Mary, marry me.

Geoff writes:

Doesn’t the young man actually mean that now may be all the time that there is?

I agree that this is the intended meaning, but I also believe that the lyric, as it stands, is capable of conveying this meaning by inverting the scope relation (between all and may). Since Geoff closed the post to comments, I sent him an email with my thoughts, which I’d like to reproduce below. (I’ve changed each occurrence of “you/your” to “Geoff/Geoff’s” so that it’s more readable as a standalone commentary.)

In regards to Geoff’s recent Language Log post, I’m tempted to say that there’s no problem: both readings are likely quite available, in theory, i.e. there is scope ambiguity, but context coerces one reading over the other. If we a consider a sentence like (1), where both the ∀ > ∃ and ∃ > ∀ readings are sensible, I’d argue it’s very ambiguous. (“∃” here stands for epistemic possibility.)

(1) These are all the letters (that) da Vinci may have written.

reading I (∀ > ∃): these (and only these) are the set of letters L such that for any letter l in L, it may be the case that da Vinci wrote l.
reading II (∃ > ∀): it may be the case that these (and only these) are the set of letters L such that for any letter l in L, da Vinci wrote l.

To paraphrase a bit, under reading I, “these letters” is a pile containing all the letters ever written by da Vinci, i.e. if he wrote a letter, it’s in this pile, as well as maybe (we don’t know) letters written by other people; under reading II, “these letters” may (we don’t know) be the pile containing all (and only) those letters written by da Vinci, i.e. no letters by anyone else.

Personally, I find (1) to be ambiguous in the way described above. If that feeling is shared by other native speakers, then it’s not surprising that “Now is all the time (that) there may be” should be equally ambiguous. It would then come down to saying that context (the song) coerces the inverse scope reading over the surface scope reading. This would be a particular instance of Geoff’s “radical underspecification” theory, but could still have come about due to his “performance” theory explanation. (I don’t think the two are mutually exclusive.)

By the way, we know that (non-modal) universal and existential operators can interact scopally in simple sentences like “Every student read some book”, but we can replicate the ambiguity in certain sentences with relative clauses, as well.

(2) I bought all the paintings that someone named Charles had painted.

reading I (∀ > ∃): for every painting P such that there was some Charles who had painted P, I bought P.
reading II (∃ > ∀): there was some Charles such that for every painting P that Charles painted, I bought P.

Under reading I, I (probably) came away with a bunch of paintings from several different Charles’s, maybe because I like the name Charles; under reading II, I came away with a bunch of paintings from one specific Charles, maybe because I like him as a painter.

Curiously, if we switch the order of the universal and existential modals, as in (3), the ambiguity is lost.

(3) I bought some painting that everyone named Charles had painted.

The only available reading is that the painting I bought was co-painted by a group of people all named Charles; it can’t be interpreted with reading I above. Thus, if an epistemic possibility modal acts like an existential operator, we expect to find the same contrast, and I think we do, in (4).

(4) The museum contains some letter that every Florentine scientist from the Renaissance wrote.

The only available reading, for me, is that there is some letter co-written by every Florentine scientist; it can’t be interpreted with the reading that for every Florentine scientist, the museum has some letter written by him.

So I think we have pretty good evidence not only that existential and universal operators can interact scopally, but also that epistemic possibility modals are much like existential operators. As such, it would make sense that the lyric from the song, just like (1), is ambiguous, and that the relevant reading emerges by contextual coercion. (Plus, the surface scope reading there is just plain weird.)

Posted in modality, scope ambiguity, semantics | Leave a comment

are you gonna leave?

Following up just briefly on the gonna stuff from the end of the previous post, I noticed the following contrast last night.

Context: it’s getting late and you’re ready to leave the party, so you tell your friend:
(1) I’m leaving.
(2) I’m gonna leave.

Context: your friend sees you putting your coat on, so she asks:
(3) Are you leaving?
(4) #Are you gonna leave?

Sentences (1) and (2) are both perfectly acceptable when leaving a party, although I’d say (1) a bit more abrupt, or slightly less polite (in some cases) than (2).

In the case of questioning someone, however, at least in this context, the gonna (or going to) version in (4) sounds odd to me. I hear a lot of non-natives use it, especially here in Montreal.

It’s certainly acceptable in other contexts, though, e.g. when uttered with a tone of impatience, because you want the person to leave a.s.a.p.

Context: at a bar, a crude man with whiskey breath, unkempt hair, and an offensively ugly Hawaiian t-shirt ignores the systematic rejections of the girl he’s trying to pick up, so finally she asks:
(5) Are you gonna leave, or what?

Or maybe francophone Montrealers are just impatient with me.

Posted in Uncategorized | 2 Comments

been supposed to for…

There are what I might, for now, unscientifically call “gaps” in language, in the sense that certain pieces of a particular pattern in a grammar are missing, and are not replaced/suppleted by anything else. In this sense, the absences of, say, goed or goodest are not gaps since they’re suppleted by went and best, respectively.

Likewise, the absence of a future or infinitive modal can, as in (3) and (4), is perhaps not a gap since we can reformulate the idea with (be) able to, as in (5) and (6), and can and (be) able to are pretty synonymous, I think.

(1) Harold can recite 1,000 digits of pi, but not necessarily in order.
(2) When he was 12, Harold could recite 1,000 digits …

(3) *By next year, Harold will can recite 1,000 digits …
(4) *Harold hopes to can recite 1,000 digits …

(5) By next year, Harold will be able to recite 1,000 digits …
(6) Harold hopes to be able to recite 1,000 digits …

An example of what I’d consider a gap, and quite a frustrating one sometimes, is the absence of many forms of be supposed to that one would possibly think ought to be okay. The one that always gets me is the past perfect, as in (9) and (10).

(7) My landlord is supposed to fix the door.
(8) My landlord was supposed to fix the door 3 weeks ago.

(9) *My landlord has been supposed to fix the door for 3 weeks.
(10) *It’s been 3 weeks since my landlord has been supposed to fix the door.

According to a very technical and precise statistical analysis of about four or five people’s responses on Facebook, as well as a short discussion on WordReference, present perfect has been supposed to is ungrammatical.1 And the same goes for the infinitive, i.e. be supposed to preceded by to or a modal, as well as for the gerund.

(11) *My landlord claimed to be supposed to go out of town.
(12) *My landlord might/could/should/must/would be supposed to go out of town.
(13) *My landlord denied being supposed to fix the door.

But there’s a bit of weirdness when it comes to the subjunctive.

(14) If my landlord is supposed to fix the door tomorrow, then he’ll come around 8:00 AM.
(15) ?If my landlord had been supposed to fix the door yesterday, then he would’ve come around 8:00 AM.

(16) If my landlord were supposed to fix the door today, then he would be here by now.
(17) *I insisted that my landlord be supposed to fix the door today.2

Considering has been supposed to in (9) and (10) is bad, it’s strange that had been supposed to in (15) is relatively better, albeit still degraded compared to (14). That is, it’s not as good as (14), but oddly better than (9) and (10). Moreover, the past subjunctive were in (16) is perfectly fine, leading one to think that the present subjunctive be is fine too, but that’s not the case. (Thanks to entangledbank for these sentences.)

The reason I consider this a gap is because there really is no semantically equivalent alternative for supposed to like there is for can. The closest in my mind is probably have to, but there’s definitely a difference in meaning between the two, whereas there is no such difference (that I can perceive) between can and be able to in most contexts.

One explanation for some of the above facts might be that be supposed to is modal not only semantically, but also syntactically, just like can. That is, it (or at least the inflected part) is generated in/occupies the same syntactic position as words like can, might, to and would, hence why they cannot co-occur.

But that doesn’t explain why has been supposed to is bad. If is and was are acceptable inflections, why not has been? And why is subjunctive had been slightly better, but not perfect? And what’s wrong with the gerund?

It’s interesting that this use of supposed to also has a particular pronunciation to it, which contrasts with supposed to in the sense of “assumed to”, as in (18).

(18) The killer is supposed (by the authorities) to have fled the country by now.

Obligation supposed to is pronounced something like [səˈpoʊstə], with voiceless [s], whereas “assumed” supposed to is [səˈpoʊzd + tə (tu)] (two distinct words), with voiced [z], which of course makes sense since it’s the past participle of the verb suppose, which has [z], not [s]. Obligation supposed is arguably no longer really the participle of anything.

In a way, then, maybe obligation be supposed to has been re-analyzed/re-lexicalized. It’s similar to gonna, which in fact seems to have the same sort of distribution.

(19) My landlord is gonna fix the door.
(20) My landlord was gonna fix the door.
(21) If my landlord were gonna fix the door, …

(22) *If my landlord had been gonna fix the door, …
(23) *I insisted that my landlord be gonna fix the door.
(24) *My landlord claimed to be gonna fix the door.
(25) *My landlord might/could/should/must would be gonna fix the door.
(26) *My landlord denied being gonna fix the door.

Some (but not all) of the sentences of above actually get better, at least to me, with the unreduced going to in place of gonna, e.g. (27).

(27) ?My landlord claimed to be going to fix the door tomorrow.

Read with a future meaning, not a movement meaning, of course. With be supposed to, however, there is no such semantically equivalent unreduced version available.

1Note that I’m disregarding, for now, the use of supposed to in the sense of “assumed to”, as in (18).
2It’s entirely possible that (17) is bad for semantic reasons: it’s rather redundant. Unfortunately, I can’t think of any verbs/adjectives that take the subjunctive that aren’t verbs/adjectives of necessity/obligation, which be supposed to already conveys.

Posted in modality, semantics, syntax | Leave a comment


I remember the first time I was assigned to write a squib, a couple years ago in a phonology course, I had no idea what I was being asked to do. The dictionary entry for squib didn’t help much.

1. a short and witty or sarcastic saying or writing.
2. Journalism a short news story, often used as a filler.
3. a small firework, consisting of a tube or ball filled with powder, that burns with a hissing noise terminated usually by a slight explosion.

Presumably, my phonology write-up wasn’t supposed to be witty or sarcastic, but what did I know? I was bit embarrassed to ask the prof or other students since it seemed squib was a word budding linguists were expected to know, so I just wrote up my analysis of Spanish spirantization (or whatever), and that was that. Since then, I’ve written lots of squibs, but only recently did I find out what a squib really is—at least in linguistics, and at least originally.

Poking around online a couple months ago, I stumbled on Squibnet, a collection of scanned, handwritten squibs by the linguist Haj Ross (as well as some by Paul Postal). I don’t know whether or not Ross was the first to commandeer squib (the OED dates squib to the 16th century) and use it with a meaning specific to language and linguistics, but in any case, he does provide a nice description of what a squib is to him.

Squibs are short notes about kinky facts of language. They may occasionally be welcome, in that they provide evidence for someone’s pet theory. Most frequently, however, they are rambunctious, insolent, nose-thumbing bazookas, taunting theoreticians of every stripe, daring them to stretch their minds enough to wrap around the damned facts the squibs call to our attention.

In Athenian Greece, poets were not allowed to be citizens – they were too unpredictable, irreverent. Squibs are the poets sneering outside the walls of Theoretopolis, mocking us.

But with luck, squibs become seeds.

If you look at a few of the scanned squibs, you’ll see that many are nothing more than a few scribbled words, arrows, and question marks. They really are just short notes of interesting ideas. Nowadays in linguistics, I think squib has come to denote something longer and more detailed, though not as long and detailed as, say, an article.

Whatever a squib is, the purpose of this blog is to get myself into the habit of jotting down ideas and thoughts, keeping productive. Topics will certainly include language and linguistics, but also things like math, logic, science, computery stuff, maybe even music and film, and whatever other random nerdery comes to mind. With enough writing, and with any luck, I’ll figure out what squib means to me.


Posted in Uncategorized | Leave a comment