Lera Boroditsky has a super-interesting essay at Edge on her work empirically testing the proposition that language structures thought. (Blërg – resisting urge to… blockquote…. sigh.)
So Boroditsky’s got some clever tests, including asking speakers/writers of a different language to arrange pictures chronologically (Roman languages tend to arrange chronology from left to right, Hebrew from right to left, and fascinatingly, the Kuuk Thaayorre in Australia do it from east to west), and testing incidences of adjectives speakers of languages with gendered nouns assign to those nouns – Germans think keys (male) are hard and jagged and bridges are slender and beautiful, where Spanish-speakers (whose gender assignations switch the nouns) correspondingly flip associations.
But… okay, look. I believe in this thesis. But the tests to my mind are not conclusive evidence. Here’s why.
You can’t get into a person’s head.
Is is that simple? It is.
Because (stay with me) all of these tests don’t show that speakers of different language think differently, but that they represent thought differently. The way we write changes the way we talk, and the way we represent thought in space. The way we talk also changes the way we write. And the way we talk changes the way we talk. You don’t have any evidence – at least, any evidence that doesn’t assume the premise – that Germans actually THINK bridges are more graceful or beautiful than Spaniards do – just that they’re more likely to use adjectives with feminine associations with feminine nouns. What this suggests immediately is that language is a complex and interconnected system where terms and kinds group together, and small linguistic changes actually trigger a series of different linguistic associations and values. It DOESN’T immediately prove that language structures thought – understood as something independent from its representation.
Because if language is the vocal and visual representation of concepts, then ALL of Boroditsky’s tests are instances of language. Language structures language. And once you assume unproblematically that language directly represents thought, then you naturally discover that thought and language are inseparable. Which is what was to be shown. But this is logically a tautology – even if its empirical specifics of how that tautology manifests itself are fascinating.
Let me reframe this, then. What I think these experiments show is that in moments where we may think we are simply registering our pure and unmediated experience of the world, we’re really on auto-pilot – language is in fact doing our “thinking” for us. But this kind of not-quite-thinking doesn’t automatically deserve to be called “thought” at all.