The Analogy Engine
Dec. 14th, 2008 03:22 pm![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
mtvesselAug 2008
The Stuff of Thought - Steven Pinker - Penguin, 2007
* * * *
I have long been suspicious of Noam Chomsky's concept of a universal grammar, the principle that all humans are born with a grammatical "toolkit" which they use to acquire and develop language. Whilst it explains some curious similarities in sentence construction between culturally distinct languages and provides a possible mechanism for the astonishing speed at which infants learn to speak, it has always seemed to me to be a bit too convenient to be true, an unverifiable just-so story. Reading this book, however, has caused me to revise my opinion. For if the universal grammar hypothesis is true, then the study of language - determining the nature of the tools in the toolkit - should give us important clues about how our minds work and, importantly, what the limitations on human cognitive abilities might be that could affect our continued survival on this planet. With the many examples collected in this book, Steven Pinker makes a persuasive case that linguistics can indeed offer these sorts of insights.
The first chapter is a case in point. It concerns the curious phenomenon of why some verbs are transitive (direct and indirect objects can be swapped with use of appropriate prepositions) and others are not. Consider the following four statements:
Jack loads hay into the wagon
Jack loads the wagon with hay
Jill fills the glass with water
Jill fills water into the glass
Three of these are grammatical English, but the fourth is not. Why? Surely both “load” and “fill” are used for the same class of action (putting something into a container), so they should follow the same grammatical rules?
Clearly there must be subtle rules that determines whether such content-locative verbs are transitive or not, and they arise, says Pinker, from the way in which the brain processes information about objects and actions. It turns out that verbs can be divided into “micro-classes” based on geometry (such as whether the content is being forced into the container or whether the content is coming out of the container) and physics (whether the agent is applying a force directly, such as spreading butter onto toast, or allowing gravity to do the work, such as dripping honey on it). Verbs that imply direct action of an agent and a mutually felt force for both container and content are transitive: those where the action of the agent is indirect, or which concern themselves only with the change of state of the container or the content, are not.
It turns out that micro-classes of verbs that follow the same grammatical rules exist in unrelated languages and that often (but not always) the classes can be distinguished on similar grounds of agency, geometry and physics. This strongly suggests that humans have a common mental model which they use to organise and categorise their sense experiences. This model is pre-linguistic; sensibly, Pinker pooh-poohs the pronouncements of the linguistic determinists such as Whorf, Wittgenstein and Nietzsche that we require language to classify concepts and hence to think, an idea which is clearly nonsensical to artists, musicians and others who seek to convey meaning in non-verbal media.
In succeeding chapters, Pinker shows that similar models underlie the linguistic behaviour of nouns relating to space, time, substances and causation; the coining of neologisms; veiled threats and other implications in speech (which can be shown to be the result of a prisoner’s dilemma-type calculation of possible payoffs); and most entertainingly, taboo concepts and swear words (rude verbs for sex are transitive, implying direct action or force, whereas polite/politically correct ones are not). This does of course raise a worrying possibility. Are we just programmed robots, forever stuck with our abstract mental models and incapable of thinking, feeling or acting beyond them? Have we simply removed determinism from language and instead applied it to the toolkit that underlies it?
Encouragingly, Pinker’s answer is no, and the reason for believing this comes from another important feature of our linguistic landscape, which is metaphor. The human brain is adept at taking patterns found in one situation and applying them (often inappropriately by any logical criteria) to another. This, combined with the combinatorial structure of language which allows a vast number of grammatically valid phrases to be constructed from a relatively small number of concepts and relational rules, allows us to express a near-infinite number of new ideas. So we are not stuck; because our brains are analogy engines, we can always come up with new patterns and new models. The toolkit of thought that nature has given us may be fixed, but the things that we can do with it are not.
The Stuff of Thought - Steven Pinker - Penguin, 2007
* * * *
I have long been suspicious of Noam Chomsky's concept of a universal grammar, the principle that all humans are born with a grammatical "toolkit" which they use to acquire and develop language. Whilst it explains some curious similarities in sentence construction between culturally distinct languages and provides a possible mechanism for the astonishing speed at which infants learn to speak, it has always seemed to me to be a bit too convenient to be true, an unverifiable just-so story. Reading this book, however, has caused me to revise my opinion. For if the universal grammar hypothesis is true, then the study of language - determining the nature of the tools in the toolkit - should give us important clues about how our minds work and, importantly, what the limitations on human cognitive abilities might be that could affect our continued survival on this planet. With the many examples collected in this book, Steven Pinker makes a persuasive case that linguistics can indeed offer these sorts of insights.
The first chapter is a case in point. It concerns the curious phenomenon of why some verbs are transitive (direct and indirect objects can be swapped with use of appropriate prepositions) and others are not. Consider the following four statements:
Jack loads hay into the wagon
Jack loads the wagon with hay
Jill fills the glass with water
Jill fills water into the glass
Three of these are grammatical English, but the fourth is not. Why? Surely both “load” and “fill” are used for the same class of action (putting something into a container), so they should follow the same grammatical rules?
Clearly there must be subtle rules that determines whether such content-locative verbs are transitive or not, and they arise, says Pinker, from the way in which the brain processes information about objects and actions. It turns out that verbs can be divided into “micro-classes” based on geometry (such as whether the content is being forced into the container or whether the content is coming out of the container) and physics (whether the agent is applying a force directly, such as spreading butter onto toast, or allowing gravity to do the work, such as dripping honey on it). Verbs that imply direct action of an agent and a mutually felt force for both container and content are transitive: those where the action of the agent is indirect, or which concern themselves only with the change of state of the container or the content, are not.
It turns out that micro-classes of verbs that follow the same grammatical rules exist in unrelated languages and that often (but not always) the classes can be distinguished on similar grounds of agency, geometry and physics. This strongly suggests that humans have a common mental model which they use to organise and categorise their sense experiences. This model is pre-linguistic; sensibly, Pinker pooh-poohs the pronouncements of the linguistic determinists such as Whorf, Wittgenstein and Nietzsche that we require language to classify concepts and hence to think, an idea which is clearly nonsensical to artists, musicians and others who seek to convey meaning in non-verbal media.
In succeeding chapters, Pinker shows that similar models underlie the linguistic behaviour of nouns relating to space, time, substances and causation; the coining of neologisms; veiled threats and other implications in speech (which can be shown to be the result of a prisoner’s dilemma-type calculation of possible payoffs); and most entertainingly, taboo concepts and swear words (rude verbs for sex are transitive, implying direct action or force, whereas polite/politically correct ones are not). This does of course raise a worrying possibility. Are we just programmed robots, forever stuck with our abstract mental models and incapable of thinking, feeling or acting beyond them? Have we simply removed determinism from language and instead applied it to the toolkit that underlies it?
Encouragingly, Pinker’s answer is no, and the reason for believing this comes from another important feature of our linguistic landscape, which is metaphor. The human brain is adept at taking patterns found in one situation and applying them (often inappropriately by any logical criteria) to another. This, combined with the combinatorial structure of language which allows a vast number of grammatically valid phrases to be constructed from a relatively small number of concepts and relational rules, allows us to express a near-infinite number of new ideas. So we are not stuck; because our brains are analogy engines, we can always come up with new patterns and new models. The toolkit of thought that nature has given us may be fixed, but the things that we can do with it are not.