Signing space
A Socratic walk-through of signing space — reasoned out one step at a time, not lectured.
The question we started with
THE QUESTION #How can a signer point at empty air and have everyone present know exactly who is meant?
Watch a signed conversation about two absent people. The signer points to a spot on her right, and everyone watching knows she means the brother rather than the doctor. There is nothing there: no object, no picture, no resemblance between the spot and the man.
The intuitive explanation is that signing is a kind of drawing in the air, so the space is a small stage on which the signer arranges her characters. That guess is worth testing rather than accepting, because it predicts something specific: the arrangement should be depictive, matching how things really are, and a viewer should be able to read it without preparation. Both halves turn out to be wrong, and finding out why is the whole of the answer.
Reasoning it through
REASONING #Start with the second half, since it is easiest to check. Can a spot be interpreted before it has been used? No. A signer who points to an empty region that has not been set up has said something incomprehensible, or has changed the subject. The spot means nothing until it has been established — the signer names or describes a referent and associates it with that region, usually by signing the noun and then indexing the location, or by producing the sign itself there. Only afterwards does pointing there refer.
That single observation reframes everything. If a spot has to be assigned before it works, it is not a picture. Pictures do not need to be introduced. What needs introducing is an arbitrary token whose meaning is fixed by an act of joint attention — which is to say, a coordination problem being solved in public, in front of everyone who will later need to read it.
Now the first half. Does the arrangement match reality? Sometimes, and it does not have to. If the brother lives north and the doctor south, the signer is free to set them up the other way round, and comprehension holds; she may use her right in one conversation and her left in the next. What matters is not where the spot is but that both parties saw it assigned and are still tracking it. There is a partial exception — a physically present referent's own location is used — but for absent referents the assignment is essentially free.
So what work do these spots then do? More than pronouns. Once two referents hold two regions, a class of verbs can be signed from one region to the other, and the direction of the movement marks who did what to whom. Reverse the movement and you have reversed subject and object without changing a single handshape. The space is not decoration around the grammar; for these verbs it is where the grammatical relations are expressed. The same established regions also let a signer shift body or gaze toward one and sign as that referent, which is how reported speech is handled.
Does the account make a falsifiable prediction? Two. First, if reference here rests on public assignment rather than depiction, then reassigning a region mid-conversation should require overt work — signers should mark the change and addressees should occasionally lose track, exactly as speakers do with an overloaded he. That is observed. Second, if the mechanism is coordination rather than resemblance, a viewer who missed the establishing move should be unable to recover the reference from the spot itself. Also observed, and painfully familiar to anyone who has joined a signed conversation late.
What would refute it? If naive non-signers could reliably read who was meant by an unestablished pointing, or if signers could not set referents up in arrangements contradicting their real-world positions, the depiction account would be back and this one finished.
The analogy
THE ANALOGY #Think of two people at a table agreeing that the salt cellar stands for the north entrance and the pepper for the south. After that, moving a finger from one to the other says something exact about a route. Nothing about the salt resembles a door. A third person arriving at that moment sees two condiments and a gesture, and has no way to recover the meaning, because the meaning was fixed by an agreement they did not witness.
the salt and pepper are physically there and stay put on their own, whereas signing-space regions are held only in the memory of the people conversing — so they can be reassigned, crowded, or lost in a way objects on a table cannot, and they carry grammatical marking such as verb direction that no arrangement of condiments does.
Clarifying the model
THE MODEL #Three things to hold onto, one of which is a live disagreement rather than a settled fact.
First, the relationship to iconicity. Signing space is sometimes used depictively — to lay out a room, a route, the arrangement of objects — and that use does resemble what it represents. The mistake is treating it as the general case: depictive and referential use are different jobs done in the same space, and pointing at empty air is the referential one, where resemblance has dropped out. This is also the point of difference from the broader question of whether sign languages are full languages, which turns on their having contrastive sublexical parameters; what is at stake here is narrower — how reference is maintained across a discourse once the vocabulary is granted.
Second, plainly: the widespread belief that sign languages are pantomime, and signing space a form of mime, is false, and this is one of the sharpest demonstrations. A system in which a location must be introduced before use, can be assigned against the facts, and carries agreement marking on verbs, is not miming. Nor is that belief a harmless simplification — it has historically underwritten the prescriptive judgement that signing is a gestural substitute to be replaced by speech wherever possible, which is a claim about what people ought to do dressed up as a claim about what signing is.
Third, the honest caveat. Whether these spatial loci are genuinely grammatical — morphemes in an agreement system — or something that is not quite grammar is actively argued. Scott Liddell's objection is that a listable grammatical category ought to have a finite inventory, and the number of usable locations is not finite: it is a continuous space, pointed at as precisely as the situation demands, which looks more like the way a speaker points at things than like the way a language conjugates a verb. Others hold that the system is agreement with a gestural component. I have described the behaviour, which both camps agree on; how to classify it, I am reporting as unresolved.
A picture of it
THE PICTURE #How to readRead top to bottom as time in one conversation, with the two middle columns standing for regions of empty space rather than people. The first four lines are setup, and nothing before them means anything — a point at either region before its assignment is uninterpretable. The two verb lines are the payoff: the identical verb, moved in opposite directions between the same regions, reverses who acted on whom, which is grammar carried by direction rather than handshape. The final pair shows the fragility a picture-in-the-air account cannot explain: reassigning an established region costs something, because what was lost was an agreement, not a drawing.
What became clearer
WHAT CLEARED #Pointing at empty air works because the air was not empty in the only sense that matters — a referent had been publicly attached to that region, in view of everyone who would later need to read it. The space is a set of coordinates jointly held in memory, not a stage on which anyone depicts anything, which is why the assignment can contradict the world, why it must be established before use, and why joining late leaves you stranded. And the language does more with those regions than speech does with pronouns: it points verbs between them and lets direction carry the grammatical relations outright.
Where to go next
ONWARD #- Whether spatial loci are best analysed as agreement morphemes or as grammar fused with pointing gesture.