that it will take a long time to settle this issue, and that in the interim the arguments that arise will be many, vigorous, and perhaps bitter, with Bloom-Au an initial engagement.
Although I take no stand on the substance of the issue, I think that the way the controversy is shaping up suggests that there may be something wrong in the way the Sapir-Whorf hypothesis has been discussed in recent years. As usually presented, the hypothesis is graduated in degree in only one way: Slobin ( 1979), for instance, distinguishes a strong form of the hypothesis, in which language structure determines thought, from a weak form, in which language merely influences thought. Although that graduation is appropriate, it leaves the distinction between language and thought as an untouched dichotomy, whereas that dichotomy especially needs to be graduated. Indeed, one of the usual points made about the hypothesis is that, in principle, it requires independent assessment of language and thought -- in particular, that "thought" be assessed other than through language, in order to be sure that any effect found is an effect of language on thought, not just an effect of language on language or a correlation between language assessments. Of course, this methodological point stems from, and reinforces, the conception of language and thought as a dichotomy. It is obvious, however, that "language" covers a wide range of cognitive phenomena, and that the distinction between "language" and " non-language" tasks is far from allor-none.
We have long understood that speakers of different languages have acquired different rule systems. We now know that speakers of different languages have acquired different comprehension processes and strategies, in addition to the different rule systems. In the case of written languages, it is obvious that these language-specific processes must include reading as well as comprehension of spoken language, as Aaronson and Ferres's data confirm. One may surmise that they extend to writing processes and strategies. Thus, language-specific cognitive processes extend over a wide range of linguistic performances, and could well extend quite far into that grey area where language and thought are hard to distinguish. It seems to me, therefore, that rather than of a hypothesis we should speak first of all of a Sapir-Whorf question: How deeply and into how wide a range of cognitive processes do language-specific processing differences extend? The Sapir-Whorf hypothesis is that the answer to that question is "Quite deeply and quite widely." Of course, the ultimate answer to the Sapir-Whorf question will be determined by research, and will be much more specific than a simple affirmation or denial of the hypothesis.
Au, T. K. ( 1983). Chinese and English counterfactuals: The Sapir-Whorf hypothesis revisited. Cognition, 15, 155-187.