I just read the part in Pinker's The Stuff of Thought where he talks about linguistic determinism.

He mentions what he calls "the current poster child for linguistic determinism", the work of Stephen Levinson and others on the effects of language on spatial cognition.

In an experiment described in Pederson et al. (1998), which you can find by following the Levinson link above, speakers of Tzeltal (a Mayan language which does not have clear words for "right" and "left" and primarily uses references to local geography to describe spatial relationships) and speakers of Dutch (which, like English, readily uses "right" and "left") were shown a configuration of toy animals lined up on a table, tail to nose in a row, facing either to the left or right of the subject.

They were asked to memorize the display, and then they were turned 180 degrees to an empty table, given the three toys and asked to "make it like it was before", an ambiguous task. The hypothesis was that speakers of Dutch would be more likely to line up the animals in a way relative to themselves, while Tzeltal speakers would be more likely to line them up facing the same way in regard to an absolute (geocentric) frame of reference. And that's what the experimenters found.

There have been a series of back and forth experiments between rival groups, trying to show that this effect is either bogus or real.

I actually wrote a paper on some of these interchanges, and I became relatively disillusioned with both groups, who seemed to stubbornly change aspects of the experiments with no good reason. For example, in the original experiments, the researchers did not control for the environment (e.g. whether or not they were carried out indoors or out, and this is a huge variable). In the rebuttal follow-ups, the researchers did not duplicate a memory load that was part of the original experiments.

Pinker takes the side of the detractors, essentially dismissing the claims of Levinson and his colleagues as bold exaggerations. I have no problem believing that language influences non-linguistic thought, since the brain is a massively interconnected hierarchy, and it would be surprising if the upper levels of the hierarchy of language works as an abstract system of lower-level concepts didn't exert some top-down influence on the working of the lower levels of the hierarchy.

The relative impact of this influence is a somewhat interesting question, but I'm with Pinker to the extent that I think it is the foundational non-linguistic thought that exerts the greater influence on language, rather than the other way around. However, I wouldn't dismiss it out of hand the way he does.

In the end the whole argument seems to generate a lot of heat and not a lot of light. The exchanges between the competing camps at times get downright ugly, and for no good reason. At one point I thought of trying to replicate the animals-in-a-row experiment, but I'm glad I didn't. It would simply have been a distraction in a debate where I think little can actually be resolved and much of the argument comes from poor assumptions and miscommunication.


[1] Pederson, E., Danziger, E., Wilkins, D., Levinson, S.C., Kita, S. & Senft, G. 1998. Semantic typology and spatial conceptualization. Language, 74(3): 557-589.