Does a martial-arts action strike you as a well-executed ballet or a confusing oleo of hands and feet?
The current thesis favored on this blog is that language is a system for directing one another's attention so that we can share perceptions, real, imaginary or metaphorical. As it stands now I propose that human evolution began with the formation of communities based on cooperation and sharing. Once our ancestor moved from social to communal arrangements the normal, individualistic, Darwinian impediments to sharing gave way to the group benefits of cooperation and trust. Language became a part of the new order in which, initially, people spoke literally, pointing out perceptible details of reality.
Having spent some years assembling the ideas in that last paragraph, I am these days mainly interested in two questions: What new evidence have you got to challenge the thesis? What new evidence have you got that supports it?
In this regard, the current issue of Language and Linguistics Compass includes an interesting paper by Eva Malaia titled, "It Still Isn't Over: Evidence boundaries in language and perception" (abstract here). The paper is a summary of twenty-first-century thought regarding event perception and its relation to language.
Perception has often been considered as a static process, such as binding sensory impressions into a whole so that we can identify an object or scene. Event perceptions concern changes over time. They too must identify objects, but now the perception must bind the objects into a dynamic scene (an event).
It turns out to be difficult to see an event we have never seen before. Thus, the moves of a martial-arts expert can be recalled by somebody unfamiliar with the moves as just a confusing bit of motion. Yet an experienced viewer cannot only see the moves, but can critique them, offering advice on how to perform the action better. Malaia offers this model of how we perceive an events:
- Events can be divided into segments, which we can define as the minimum perceivable part of an event. Segments are individuated and limited in time.
- Semantic memory holds a series of templates (schema) that define event segments.
- As the senses provide data, it is matched against the templates.
- If a template matches, working memory is freed up for remembering details of the event segment.
This simple process explains the difference in the martial-arts observers. The naïve observer has no template, makes no match, stresses working memory and has no clear recollection of the segments comprising the event. Meanwhile, the expert observer has a set of templates associated with the event, matches segments, and frees up working memory to notice details about a segment.
It would be fascinating to know how important language is for breaking a complex action down into segments. A teacher can show and identify the segment of an event so that students take note of them and see them. If I watched a film of a martial arts event twenty times, just watched without an instructor, how much clearer would the event be to me after the 20th viewing? I suspect it would be better just because I've talked to myself about what I'm seeing. If a chimpanzee could be persuaded to view the same footage twenty times, would it too be able to see the event more clearly (i.e., in more segmented form).
Malaia does report that event segments "found in most of the world's languages relate closely to the current understanding of how the brain processes reality." [p. 90]
- "Speakers of most known languages describe their reality as a collection of separate structured events." [p. 90] So we can say, She was running along when she tripped over a stone and fell down hard. I was worried she might be badly hurt, but she bounced back up laughing. What that passage does, of course, is break one event down into a series of segments. Notice by the way that each segment has its own verb: was running…tripped over… fell down … was worried…might be hurt... bounced back up.
- "In general… humans [of whatever culture] rely on visual cues of velocity and acceleration to separate events in their surroundings, and do so with remarkable similarity." [p. 91] Three of the verbs mentioned above seem to report a constant velocity: was running [literal velocity] … was worried [metaphorical constancy]...might be hurt [potential constancy]. The other verbs—tripped over … fell down … bounced back up—report a change in velocity. Notice anything odd? The constant actions use an auxiliary verb while the actions that change speed include prepositions. In other words, the kind of segment perceived can be expressed in the verbs. Of course, it is possible to omit all that: She ran, tripped, and fell hard. I worried she hurt herself, but she rose laughing. So the specification of segment type is not obligatory, although I cannot help noticing that the second version is much less visual and interesting. Before reading Malaia's paper I could not have explained so neatly why the first version was preferable, although my ear would have told me it was.
The paper also reports experiments showing that the same kind of working-memory advantages which come from perceiving events in segments, come from reading about events in segments.
Whether reading or watching, part of the task is register and remember the input. It seems that we remember what we witness by breaking an event into segments, but this solution requires doing two things at once. We have to match perception with templates and be able to recall the appropriate templates later. It is this two-part operation that makes it so hard for a novice observer to understand a martial-arts maneuver. Without much in the way of a matching template, there is not much ability to observe and remember at the same time.
Event and sentence segments are "entirely congruent" [p. 92]. That is to say, language breaks up events in the same way perception does and for the same reason. A reader needs to register what is being read and be able to remember it as well.
Malaia's paper does not make this point, but I suspect that her summary has an interesting implication for linguists. It is well known that the sentences of natural languages are unbounded. That is to say they can be of an unpredictable length because the rules for adding phrases can be repeated indefinitely: e.g., Bob, all dressed in white, came to the party. Bob, all dressed in white, and Mary, wearing a pink apron, came to the party. Bob, all dressed in white, Mary, wearing a pink apron, and Jack, sporting cowboy boots, came to the party…
This kind of recursive structure seems to follow naturally from the structure of event perception. The more segmented the perceptual templates, the more that can be perceived, and if language is a means of sharing perceptions it should also allow for increasingly segmented linguistic templates.
Chalk one up for the search for evidence supporting this blog's claim.