Sunday, July 1, 2007

Language, Writing, and the Spatial Representation of Events

Picture in your head one person throwing a ball to another. How were the two people oriented spatially? Was one on the left, and the other on the right? If so, which one was on the left, and which on the right? Chances are, the thrower was on the left, and the catcher was on the right. For some reason, that seems to be our default way of representing actions: with the actor on the right, the patient on the left, and the actions occurring from left to right(1), as in this beautifully drawn figure:


Why is that? Good question, but before we answer it, let's look at some research.

The most striking work on the relationship between language and the spatial representation of actions has been done by Chatterjee and his colleagues. They first looked at the relationship in a patient with profound agrammatism, an aphasia that makes it difficult to put grammatical sentences together. Here's an example of their agrrammatic aphasic's speech(2, p. 58):

Well, uh, essentially language abandon preposition. I telegraph... I, I... consciously, uh, continuity...I, I, uh, this subtle of prepositional phrases this simply cannot do. Under stress, under stress rapid I just flustered ... but continue to do basically.

As you can see, it's a mess. When they had this aphasic identify the agents and patients in pictures, he almost always said that the figure on the left was the agent, and the one on the right was the patient, regardless of which one was actually the agent and the patient in the figure(3).

Interesting, but this could be the product of the aphasia right? What about people with normal speech and comprehension abilities? To answer this question, Chatterjee et al.(4) conducted a series of studies with non-aphasic participants. In their first study, they had participants (all right-handed) read one of three types of sentences: an action by a patient in the direction of the patient (e.g., "John pushes Tom"), an action by an agent towards the agent and away from the patient ("John pulls Tom"), or a state requiring an agent and a patient ("John likes Tom"). Participants were then asked to draw the events on a piece of paper. In almost all cases, participants drew the agents on the left and patients on the right for each type of sentence, with drawings of the first type (action moving from agent towards the patient, as in "John pushes Tom") tending to represent the agent the furthest to the left relative to the patient.

In their second experiment, they had participants draw actions that occurred on either a horizontal (e.g., "staggering drunk") or vertical (e.g., "falling book") axis. The vertical actions were used as a control. For the horizontal actions, participants drew 8 out of 10 moving from left to right, on average. Here are a couple example drawings, from their Figure 1 (p. 398):


In their third experiment, Chatterjee et al. had participants listen to sentences, after which they were presented with figures that either represented the actions in the sentences (e.g., "Square pushes Circle," with a picture of a square-headed stick figure pushing a circle-headed stick figure). The sentences either involved actions moving from the actor to the patient (like pushing) or from the patient to the agent (like pulling). Half of the pictures presented the agent on the left and the patient on the right. The participants were told to indicate as quickly as they could (by pressing the right or left mouse buttons) whether the picture represented the action in the sentence they'd just heard. Overall, participants were faster to indicate whether the picture represented the sentence when the agent was on the left and the patient on the right.

So there's good evidence that we represent actors on the left and agents on the right, and tend to think of actions as occurring from left to right. But why? Chatterjee et al. argue that this is because our representations of actions have an inherent spatial component, and that this is due to the way our nervous system is organized. They write:

The left to right directional bias is likely to be linked to the neural encoding of events[ Both cerebral hemispheres deploy spatial attention with vectors in opposing directions[ The left hemisphere deploys spatial attention with a vector from left to right. As previously speculated, development of languagein the left hemisphere may have coopted left hemisphere spatial attentional networks opportunistically. An overlap of neural circuits mediating spatial attention, the directional representations of events and the instantiation of verbs, may provide the neural link between the spatial and propositional representation of events. (p. 401)

However, recent evidence argues against this explanation. Several studies have shown that adults who learned to write in a right-to-left writing system (as in Hebrew), as opposed to left-to-right (as in English), tend to put agents on the right and patients on the left, with actions tending to be represented as moving from right to left. In other words, the inherent spatial aspect of action representations could be a product of the writing system we use, rather than the wiring of our brain.


inês tello said...

My name is Inês, I´m a speech therapist from Portugal and i really enjoyed your blog. Congratulations!

inês tello said...

My name is Inês, I´m a speech therapist from Portugal and i really enjoyed your blog. Congratulations!