Let’s Come Somewhere Else

Define “come” and “go” to yourself. Go ahead, say it out loud — I won’t comment on how funny your voice sounds. Does your definition look something like this?

Interesting. Well you’re wrong (and your voice does sound a little funny).

Most people usually think of ingress and egress — moving either towards or away from the speaker. And in some ways that’s true!

1. Come towards me!
2. Come away from me!
3. Go towards me! ¹
4. Go away from me!

The problem is sentences like this:

5. I’m coming over to your house.

If (5) is said over the phone, with the listener sitting in their own home, it works just fine. But if both speaker and listener are at the park talking, (5) no longer makes sense! It’s almost as if the speaker is taking the point of view of the listener, thinking “what verb would they use?”, and conforming to that.

Is this just a psychology thing? Well… not all languages work this way, so not exactly.

To me, the conclusion is that the English words “come” and “go” have more complicated semantics than ingress or egress. With a little bit of trial and error, we can pretty succinctly describe those rules. Before we do that, though, let’s take this opportunity to learn a new word!

Deixis

Deixis

Language’s ability to “point” to the world in context.
A word is deictic if its semantic meaning is fixed, but its denotation depends on the context.
i.e. “this” and “that” are the linguistic equivalents to (and often accompanied by) physical pointing.

This dips heavily into the forbidden art of psychology. But, even though deixis gets messy, there is a sense that the kind of deixis available to us is more or less universal.

Typically deictic expressions are oriented around the deictic center. Along all kinds of axes like person, time, and location, there is a center, and points radiating outward.

This picture isn’t exactly complete, though! For example, in Japanese there is actually a third outer-ring missing from the picture above!

That is あそこ is a version of there meaning “neither location of speaker nor hearer”. And three levels isn’t the limit — Sinhala (spoken in Sri Lanka) has four, distinguishing between a “visible” and “out-of-sight” 3rd person. And, I kid you not, the Canadian Inuit language has claim to fourteen different spatial deictic terms (Denny 1982)!

Even the axes of deixis vary! Honorifics are a kind of deixis encoding the difference in social status between speaker and listener — common in many languages, totally absent in others. The direction of motion is sometimes also captured, as is the case with the German hin/her prefixes (related to the archaic “hither”/“thither” in English).

But we’re always on this hunt for a universal picture. Some theorize that all deixis boils down to a foreground/background mechanic. “Foregrounded” items are near, salient, or recent, whereas “backgrounded” are far, oblique, or past. I find that interesting when you think about axis of information.

Consider this: “this” can refer to things you’re about to say. And if you think that was interesting, “that” can refer to things you’ve already said.

Obviously there is a lot to say about deixis, so I would highly recommend Spatial Deixis-The Use of Spatial Co-ordinates in Spoken Language by Barbara Cairns -Cairns (1991) — it’s just a good introduction on the subject, full of surprising uses of language.

Formally Encoding Deixis

Typically deictic expressions aren’t actually 1-to-1 with denotations, and act more like “filters”. Take English pronouns as an example:

\(\cont\) here is the “context”, where we lookup information. We’ll use this idea of constraints to express come and go.

The Real Rules of Coming and Going

To find the real rules governing “come” and “go”, basically I wrote down a lot of sentences. I started with a template sentence:

6. I am coming/going to New York!

Then I tweaked a bunch of parameters, like person, tense, who was and was not in New York, and looked at which made sense.

In the present tense, we can look at the results as a bunch of Venn diagrams. Each circle of the Venn Diagram shows if someone is or isn’t in New York. We have a “Speaker” circle, “Hearer” circle, and maybe a “3rd Person” circle.

Now we can see that “I am coming to New York” works if the listener is in New York. “I am going to New York” works as long as I am not already in New York (though using come is often a bit more natural).

Formalization

Formally, our naïve come and go definition looked something like this²:

Now we want to take the data above, and make some new rules! I should note, these rules are going to look a lot like (though are a slight improvement on) those developed by Fillmore (1966). Looking at the Venn diagrams, there’s one pretty obvious rule³:

You can’t go somewhere you already are! Next, there is a clear rule governing go: the speaker cannot be present at the goal.

The rule for come is a little trickier: either speaker or hearer must be present at the goal (this matches the point-of-view intuitions).

Ok. Cool. … … Are we done?

What About Tense?

Ok, I kinda left something out here. It turns out tense matters when using come or go! Consider this little example:

In this example, neither speaker or hearer are in New York, but come was used! It seemed to matter more that the hearer would be in New York when the action was going to take place. But then there is this example:

Now neither participant was present at action time!

The Final Rules

We can clean up our rules by adding this temporal constraint on things. I’ll use “\(@ \text{time}\)” to mean “true at a time”. Gluing everything together gives us these rules:

Parting Notes

I found these rules pretty surprising! They’re far more complicated than what one would think for two words as “simple” as come and go!

I’ve been spending a lot of time thinking about these words, their rules, exceptions and edge cases, and have been really fascinated by their intricacies — especially cross-linguistically! For the sake of time, though, let me just leave you with some teasers:

• These rules seem to hold for the words “bring” and “take”.
• I said the speaker “takes the point of view of the listener” — there is actually psychological data for this! For example, children with Autism seem to use come and go differently than other children (Hobson, García-Pérez, and Lee 2010).
• These rules don’t seem to be exactly complete. “I’ll come with!” is one example that confuses me. Interrogatives and modals seem to also have odd behaviors.

I’ll leave you all with some further reading, if anyone is interested!

• This post is a micro version of a paper I wrote for a semantics course.
• Again, Cairns (1991) is a great introduction to Deixis, and a thoroughly enjoyable read.
• Fillmore (1966) is the paper on “Come” and “Go”.
• Wilkins and Hill (1995) is a paper on “Come” and “Go” in two very different languages. The paper challenges just how “basic” these words really are.

References