When I began this blog I agreed with the Sanskrit poet who said the greatest wonder of language is its ability to make public what is in our hearts. Work on this blog, however, has made me consider a second candidate for wonder: the transitive clause. For those who slept through that grammar lesson, a transitive clause combines two things into a single action: e.g., The lion stalked the zebra; I ate the pie; The arc of history bends toward justice. Some animals can voice emotions, but no other animal or mathematical communication system can use verbs to unite subjects with objects.
Equally remarkable is the fact that all natural languages allow for transitive clauses. There are languages that do not distinguish between male and female pronouns, and there are languages that pluralize trees in a different way than they pluralize books, but every language gives its speakers a way to say I ate the pie.
Yet languages do not require that transitive clauses have the same structure. I in English is called the subject (S). It could also be called the agent or doer. Meanwhile, the pie is called the object (O) or direct object or patient. Finally, there is a verb (V), ate. The standard English structure of a transitive clause is SVO for [I] [ate] [the pie].
Mathematically, there are 5 other ways the clause could be structured:
- SOV – I the pie ate.
- OVS – The pie ate I.
- OSV – The pie I ate.
- VSO – Ate I the pie.
- VOS – Ate the pie I.
Are any of these structures more natural than others?
A recent paper by Irit Meir and 9 other authors investigates the issue of the structure of transitive clauses (“The Effect of Being Human and the Basis of Grammatical Word Order,” Cognition, abstract here). They report that out of a survey of 1,377 languages, 85% (or 1,186) have a dominant structure in their transitive clauses. The distribution of structures Is:
- SOV – I the pie ate; 565 (48%).
- SVO – I ate the pie; 488 (41%).
- VSO – Ate I the pie; 95 (8%).
- OVS – The pie ate I
- OSV – The pie I ate.
- VOS – Ate the pie I.
Every possible combination is found in some language or other, but 97% of those with a dominant pattern normally place the subject before the object. These observations have led many linguists to propose that it is natural to place the subject first. That finding gives extra support for the idea that grammatical rules, from the beginning, shape language form. Meir and the other authors have challenged this position by having experiments in which communicators invent sign or spoken languages to describe the actions observed in short video clips. They found that they get more varied structures when both the subject and object are humans (e.g., Harry cheated Donald). If the object is inanimate (e.g., pie) and the subject is human, clause structure is more fixed.
This finding seems to contradict commonsense. Normally, we expect a human to do something to an inanimate object, so even if I say The pie ate I, you can guess that it was the pie that got eaten. Speakers would seem to have more rhetorical freedom to fiddle with grammatical relations about pie than if I said Nancy punched Jane. Without some kind of settled convention, the listener will have a hard time sorting puncher from punchee. Yet this contradiction is exactly what the authors found.
The authors argue, therefore, that investigators are mistaken when they look for grammatical explanations for the structure of transitive clauses. They conclude that “the main factor at work [in the experiments] is the conceptual salience of the participants” (p. 203). The what?
Conceptual salience refers to how likely something is to grab our attention. If you are looking at a scene, there is some sort of hierarchy of objects most likely to catch your attention; in most cases, the hierarchy is likely to be people first, then animals, then moving inanimate objects, then still objects. Thus, if you are looking at a scene out west you might first notice the cowboys, then the cattle, then the tumbling tumble weeds, then the rocks. Thus, if you see Harry cheat Donald, there is less of an automatic point to grab your attention. If Donald catches your eye first, you might report Donald was cheated by Harry or Harry might grab your attention and you name him first. There will be variety in your choice of what to name first. But if you see Harry eat a pie, the human Harry is much more likely to dominate your attention and come first in your report on what happened.
Regulars on this blog will know why the study is so attractive. Language, it says here, is a product of attention. Words get their meaning by piloting attention. Syntax gets its structure by shifting attention. And now we have evidence that the speaker’s rhetorical choices also reflect a hierarchy of attention. Of course, I’m pleased. At the same time, I do not want to be carried away. Lately, psychology has been plagued by a series of unreproducible findings, so it would be nice to have other researchers confirm the basic observations of these authors. Also, it would be interesting to discover if there are predictable deviations in the language of people with different hierarchies of attention. For example, many people with autism pay more attention to inanimate things than to humans. Does their use of transitive clauses reflect this difference?