Beware of the Train

Fascinating! I'd always assumed that the strong form as espoused by Orwell (that thoughts not expressible in the language are unthinkable) was obviously false (else how would new concepts be invented, and how would languages change?), but that some appropriately weak form must be true (as evidenced by my experience with programming languages). I wonder how widespread the anti-SWH stance you describe is outside Stockholm? I see that Wikipedia largely agrees with me, but that article is listed as disputed and lacks many citations. It also raises the point that the causal relationship is not clear even in cases where features of the language seem to correlate with abilities of the people.

The sci.linguistics FAQ says the following:

According to the Sapir-Whorf hypothesis, language determines the categories and much of the content of thought. "We dissect nature along lines laid down by our native languages... We cannot talk at all except by subscribing to the organization and classification of data which the [speech community] decrees," said Whorf, in Language, Thought, and Reality (1956). "The fact of the matter is that the 'real world' is to a large extent unconsciously built up on the language habits of the group," said Sapir.

Both were students of Amerindian languages, and were drawn to this conclusion by analysis of the grammatical categories and semantic distinctions found in these languages, fascinatingly different from those found in European ones. (Neither linguist used the term 'Sapir-Whorf hypothesis', however; Whorf referred to the 'linguistic relativity principle'. Moreover, the principle was almost entirely elaborated by Whorf alone.)

The idea enjoyed a certain vogue midcentury, not only among linguists but among anthropologists, psychologists, and science fiction writers.

However, the strong form of the hypothesis is not now widely believed. The conceptual systems of one language, after all, can be explained and understood by speakers of another. And grammatical categories do not really explain cultural systems very well. Indo-European languages make gender a grammatical category, and their speakers may be sexist-- but speakers of Turkish or Chinese, languages without grammatical gender, are not notably less sexist.

Whorf's analysis of what he called "Standard Average European" languages is also questionable. E.g. he claims that "the three-tense system of SAE verbs colors all our thinking about time." Only English doesn't have three tenses; it has two, past and present; future events are expressed by the present ("I see him tomorrow"), or by a modal expression, merely one of a large class of such synthetic expressions. And for that matter, English distinguishes more like six than three times ("I had gone, I went, I just arrived, I'm going, I'm about to go, I'll go").

To prove his point, Whorf collected stories of confusions brought about by language. For instance, a man threw a spent match into what looked like a pool of water; only there was decomposing waste in the water, and escaping gas was ignited by the spark-- boom! But it's not clear that any linguistic act is involved here. The man could think the pool looked like water without thinking of the word 'water'; and he could fail to notice the flammable vapors without doing any thinking at all.

A weak form of the Sapir-Whorf hypothesis-- that language influences without determining our categories of thought-- still seems reasonable, and is even backed up by some psychological experiments-- e.g. Kay & Kempton's finding that, in distinguishing color triads, a pair distinguished by color names can seem more distinct than a pair with the 'same' name which are actually more divergent optically (American Anthropologist, March 1984).

It should be emphasized that, in their willingness to consider the idea that non-Western people have languages and worldviews that match the European's in precision and elegance, Sapir and Whorf were far ahead of their time.