From the ALWS archives: A selection of papers from the International Wittgenstein Symposia in Kirchberg am Wechsel

1. The proposal

In this paper we would like to rethink the relation between language, thought and reality once the linguistic relativity hypothesis – the idea that different languages could exert characteristic and demonstrable influences on thought¹ – has been completely abandoned. Our idea is that, once the hypothesis has been abandoned, it remains possible to theoretically pursuit the relations between language diversity and people’s ways of thinking and behaving. The relation between specific language structures and specific patterns of thinking and behaving would then be much like the relation between an excavation site and archaeological findings buried underneath it.

In attempting to show that good sense can be made of the suggestion our strategy will be as follows. First we will discuss the grounds on which the linguistic relativity hypothesis fails; the first part of the discussion will formulate an argument and the second part will generalize the argument in a discussion of two empirical approaches intended to support the relativity issue. Once this is done, we will try to offer a conciliatory position showing how the question as to relations between language and thought-cum-practical reality might be posed anew.

2. An underlying assumption of the linguistic relativity hypothesis

Any attempt to make sense of the claim that different languages influence thought in important and distinctive ways relies on acceptance of a principled distinction to be established between language and thought (Lucy 1997, p. 306; see also p. 295). If this is true, then from the very outset the proponents of linguistic relativity seem to be leaving aside the possibility that no clear line can be drawn between language and thought. In what follows we will give serious consideration to this possibility.

A principled distinction between language and thought demands that the concept of thought be theoretically kept inside definitional clothing. Scientists propose many different and differently structured grounds on which definitions of this sort might plausibly be attempted.

On the strongest proposal, that could be labeled ‘radical linguistic determinism’, relativity is sustained in terms of an identity relation between language and thought. Here the need to provide definitional clothing for the concept of thought is somehow trivialized: thought is defined as indiscernible from language from the outset. The problem with this approach can be stated in the following argument. If a way of speaking is the only evidence we have for claiming that some people think differently from ourselves, then at the end of the day we have no evidence at all: for then the criteria lack with which to sort the notion of ‘thinking differently’ out from that of ‘speaking differently’. And, under the basic assumption of radical linguistic determinism, one can always contend that the linguistic differences brought up to bear on the relativity hypothesis amount to no more than differences in people’s ways ofspeaking.

After discarding radical determinism, other possibilities of keeping thought definitionally apart from language remain to be considered. In the next section we shall consider two empirical approaches concerned with stating and supporting the hypothesis. Since even the weakest of these approaches proves misleading, serious consideration must be given to the idea that no theoretical boundary between thought and language can be drawn.

3. Critical remarks

The general form of our argument is: any attempt to definitionally enclose the concept of thought results in failure to make proper sense of the notion of ‘thinking differently’. Therefore, the underlying assumption necessary for sustaining the hypothesis of linguistic relativity proves to be the very assumption that renders the hypothesis ineffective. But let us turn now to some concrete developments.

Boroditsky et al. 2003 attempts to offer empirical support for a strong version of the relativity hypothesis by posing the following concern: “Does talking about inanimate objects as if they were masculine or feminine actually lead people to think of inanimate objects as having a gender?” (p. 68). A series of experimental results are then intended to support an affirmative answer. What the authors do not present, though, is a sufficient discussion of the following question: what does it mean for someone to think of an inanimate object as having a gender? For an answer to this question obviously bear directly on the truth of the author’s claims.

It lies beyond doubt that those scientists presented interesting results systematically relating grammatical patterns of one’s language and one’s dispositions to make certain kinds of mental association. These results, interesting in themselves as they are, still shed no light on the main question of thought – since the accurate way of describing them would have to run somewhat like: “According to experimental results obtained so far, talking about inanimate objects as if they were masculine or feminine actually lead people to think of inanimate objects as associated with a gender”. But it is obvious enough that to think of something as associated with a gender is different from thinking of it as having a gender. People might think the same about objects (e.g. that they are genderless) while making different mental associations with them. So it seems plainly incorrect to claim that people think differently about objects at all just because objects make them (however systematically) think about different things.

A rather different approach is presented in Slobin 1996. Recognizing difficulties within traditional deterministic views, Slobin proposed to investigate how different languages could differently affect that specific part of the process of thinking which is directed at “formulating an utterance” (p. 71). Thus it seemed plausible to the author that, even if linguistic diversity had no major consequences for thought in general, it should have fairly specific detectable effects on a process of “thinking for speaking”.

What should be questioned here is the plausibility of applying the concept of thought to the process labeled by the author as ‘thinking for speaking’. If someone is embarrassed because of a question, perhaps we may correctly say that she is thinking for speaking. But our statement may be correct only because it does make sense to say, in a whole range of different situations, that someone is speaking without thinking. Obviously enough, what Slobin envisages is a quite different thing. He wants to use the concept of thought in such a way as to make it reasonable that we are thinking for speaking whenever we are speaking. But then a difficulty quite similar to the one faced by linguistic determinism arises: Slobin’s usage does not provide criteria for sorting the notion of ‘thinking for speaking’ out from that of ‘speaking’. The result is that we can only regard speaking in different languages as connected with different thought processes because of the trivial fact that different speech processes are involved.

The above discussed approaches illustrate a common difficulty faced by cognitive scientists of whichever persuasions: if they insist on keeping thought inside definitional clothing they run the constant risk of making experimental results irrelevant to the issue experimentation is supposed to throw light on. The whole problem is summed up in Wittgenstein’s remark that “An ‘inner process’ stands in need of outward criteria” (2001, p. 129). The intended necessity is not that we should stipulate outward criteria for inner processes; our stipulations would arise only from neglect of the point brought about by the remark. On the contrary, we can only be sure that a given inner process is what it is and not something else if we rely on the pre-given criteria according to which that process can be publicly recognized by anyone of us.

4. A conciliatory position: retrieving culture from language

All difficulties exposed in the last sections result, in one way or another, from failure to attend to the uses of the concept of thought in everyday situations. We think a proper understanding of Wittgenstein’s idea that language reflects reality only because it is part of reality helps to relieve a good deal of our theoretical drive towards the question of language and thought.

Language can be pictured as a set of tools appropriate to the practical contexts of involvement in which they are applied. These tools do not relate to reality or thought by mirroring the latter. Nevertheless, such an “instrumental” conception of language can still be justly charged of insufficiency. It seems to make too little of the fact that, if language is indeed like a set of tools, these tools still happen to have a very typical character that must somehow be accounted for: they have their own inner laws of composition. The inner structure of languages, and specially the patterns that force us to convey certain kinds of information, is probably the main linguistic fact whose force lead people to explore the relativity hypothesis.

Here we would like to propose a conciliatory position. Language can be seen from the instrumental perspective without exclusion of the possibility that linguistic diversity be explored on the theoretical level: we just need to concede that the inner structure of our linguistic tools could suitably adapt itself to the situations in which the tools are repeatedly employed.

We cannot help thinking of the causative verb forms in Hindi as related to the deeply-ingrained structure of castes existing in Indian society; or of cardinal coordinates employed for spatial orientation as connected with people’s need for precision in outdoor handlings. These strongly intuitive observations have been traditionally rejected by researchers concerned with linguistic relativity, but they turn quite plausible once their leading hypothesis is rejected. For asserting that causative verb forms in a language could determine the existence of a given social structure (by determining people’s “ways of thinking”) is certainly much more doubtful than considering those verb forms to be part of a process by which the structure of language adapted itself to the necessities presented by a given social situation (and that, in so adapting itself, it probably improved language’s contribution to the effectivity of that form of social organization).

Finally, some might say that our suggestion is bound to leave too many interesting facts about distinct linguistic structures out of the picture; but then again no one ever thought that linguistic relativity research would succeed in finding a corresponding effect for each interesting feature about language structure. The main attractiveness of our conciliatory idea lies in the fact that it finds a place (however timid) for an account of differences among inner linguistic structures without conflicting with the outer use-bound nature of language.