Abstract: Dan Everett’s attempts to refute Quine’s Indeterminacy of Translation argument are marred by misunderstanding the nature of Quine’s project. Everett’s examples of the Indeterminacy of Translation are of a different logical type than Quine’s examples. For this reason Everett’s arguments against Quine are aiming at the wrong target. Furthermore Everett’s attempts to overcome the Indeterminacy of Translation by appeal to innate perceptual constants is at odds with the overall thesis of his book which argues against appealing to innate apparatus unless you have an adequate evolutionary account of how these innate apparatus evolved. It will be demonstrated that Everett is guilty of inconsistency in the sceptical approach he takes postulations of things like a language instinct, a morality instinct etc and the casual manner in which he argues for innate perceptual constants.

Key words: Indeterminacy of Translation, Inscrutability of Reference, Innateness.

                                                          INTRODUCTION

In his 2016 book ‘Dark Matter of the Mind: The Culturally Articulated Unconscious’[1], Dan Everett discussed and criticised Quine’s Indeterminacy of Translation Argument[2]. Quine’s IDT has spawned a large literature in both scientific circles and philosophical circles. In scientific circles the IDT has typically been used as a tool to motivate nativism about concepts. Thus cognitive scientists typically (see Pinker 1995, Markman 1989 etc), relocate the problem of radical translation into a problem of underdetermination facing the child as he learns his first language. They argue that this underdetermination is overcome because of innate constraints that limit the type of interpretation that children can make to sensory experience as they interact with others and the world. One difficulty of this approach is that (1) Assuming the existence of innate constraints is question-begging against Quine’s position (2) Even if such constraints did exist they would not necessarily rule out the IDT unless it could be guaranteed that these constraints ruled out conceptual change in adults[3] (King, 2016a). Everett’s discussion of the IDT however is different than the standard treatment in cognitive science because his discussion centres on actual linguistic practice in translating foreign languages in the field.

To some degree Everett agrees with Quine that language involves a certain amount of indeterminacy. He argues that a complete translation between two languages is never possible because of different background assumptions (sometimes ineffable), which the two languages will have. He calls these background assumptions our Dark Matter and defines them as follows:

“Dark matter of the mind is any knowledge-how or knowledge-that that is unspoken in normal circumstances, usually unarticulated even to ourselves. It may be but is not necessarily, ineffable. It emerges from acting, “languaging” and “culturing” as we learn new conventions and knowledge organization, and adopt value properties and orderings. It is shared and it is personal. It comes via emicization, apperceptions, and memory and thereby produces our sense of “self”. ( DMM p. 1)

Everett’s view of language is that it is as a cultural tool, and much of the cultural shaping of our linguistic practices is implicit, and some of it ineffable. At a cursory glance there is much that Quine and Everett have in common. Both Quine and Everett emphasise the social nature of language, both of them emphasise the importance of background beliefs in shaping our overall theory of the world (though Quine says little about ineffable knowledge), and both agree that a picture of translation involving two different languages sharing identical propositional content is wrongheaded.

Nonetheless, despite agreeing on some key issues Everett does have some difficulties with Quine’s IDT. Everett critiques Quine’s IDT based on the fact that in his many years of field work he has never come across examples that Quine mentions in his IDT. Everett notes the following:

“On the side of mistakes never made, however, Quine’s gavagai problem is one. In my field research on more than twenty languages- many of which involved monolingual situations ( D. Everett 2001; Sakel and Everett 2012), whenever I pointed at an object or asked “What’s that?” I always got an answer for an entire object. Seeing me point at a bird no one ever responded “feathers”. When asked about a manatee, no one ever answered “manatee soul. On inquiring about a child, I always got  “child”, “boy”, or “girl”, never “short hair”. (Ibid p. 267)

As a field linguist with years of experience translating other cultures languages, Everett’s views are extremely important in evaluating Quine’s radical translation thought experiment. Everett’s examples though are very different from the examples Quine gave, and the Everett’s examples crucially are not of the same logical status as Quine’s. Everett notes that every time he pointed at an object and said “what’s that?” he always got an answer that named an entire object. His first answer is that in pointing to a bird and saying “what’s that?” he never got an answer such as “feathers”. Now this is an interesting fact but it doesn’t refute Quine’s IDT. Quine acknowledges that there are many different ways a translator could go wrong in interpreting what the native he is conversing with is saying. And he discusses ways in which these mistakes could be overcome through future evidence. Thus in ‘Word and Object’ Quine discusses how even when using as precisely honed a tool as matching stimulus meanings; collateral information could result in a field linguist erroneously thinking that ‘Gavagai’ and ‘Rabbit’ were synonymous, when this is not actually the case. Quine imagines the following case:

“There may be a local rabbit-fly, unknown to the linguist, and recognizable some way off by its long wings and erratic movements; and seeing such a fly in the neighbourhood of an ill glimpsed animal could help the native to recognize the latter as a rabbit. Occular irradiations combining poor glimpses of rabbits with good ones of rabbit-flies would belong to the stimulus meaning of ‘Gavagai’ for natives generally, but not to that of ‘Rabbit’ for the linguist.” (Quine: Word and Object p. 37)

Now Quine is quick to note that even in this case the field linguist wouldn’t be fooled for too long. As the field linguist learns more and more about the language; questions more natives, interacts with them in their shared environment the linguist would overcome such mistranslations. The ‘feather’, and ‘bird’ example that Everett gives is of the same logical status as the rabbit fly example. In Everett’s exposition he speaks of himself pointing at an bird and asking “what is that?” and he notes that no native ever answered ‘feather’ instead of ‘bird’. To describe this situation in Quinean terms we need to do so in terms of how the linguist would translate what the native says when the linguist points to the bird. So suppose Everett points to a bird and the native says ‘Havagon’, based on commonsense folk-psychology Everett would surely be right to translate what the native is saying as meaning ‘bird’ and not ‘feather’. Translating what the native said as meaning ‘bird’ in this instance would presumably be verified as the field linguist interacted with more and more natives. But suppose for some strange reason the field linguist translated what the native said as meaning ‘feathers’, if this was the incorrect translation the incorrectness would be discovered pretty quickly as the word was used in various different contexts and combinations. Quine would have no difficulty with this situation which is on a par with the rabbit fly example. But of course examples like ‘rabbit fly’ and ‘feathers vs birds’ are not the cases Quine was thinking of in his IDT.

The most important aspect of Quine’s IDT is his inscrutability of reference argument[4]. Quine explicates the ISR as follows:

“Does it seem that the imagined indecision between rabbits, stages of rabbits, integral parts of rabbits, the rabbit fusion, and rabbithood must be due to some general formulation of stimulus meaning, and that it should be resoluble by a little supplementary pointing and questioning? Consider, then, how. Point to a rabbit and you have pointed to a stage of a rabbit, to an integral part of a rabbit, to the rabbit fusion, and to where rabbithood manifested. Point to an integral part of a rabbit and you have pointed again to the remaining four sorts of things; and so on around.” (ibid p. 53)

Here we can see a clear difference between Everett’s example of a bird and feathers and Quine’s examples. Quine does speak of parts of a rabbit but he is speaking of integral parts. In terms of stimulus meaning of ‘birds’ and ‘feathers’, they will diverge in easily detectable ways. The native will not assent to the word ‘bird’ if the linguist is pointing to feathers on the ground or feathers in a hat. Things are different when it comes to integral parts of rabbits; given that the part is undetached, every time you point to a rabbit you are pointing to the undetached rabbit part. There is no behavioural evidence that can decide the issue. So clearly it can be seen that Quine’s examples are on a different logical level than the examples Everett uses.

When you use Quine’s actual examples and not Everett’s replacements things look very different. From a Quinean perspective the reason translators do not find the various different alternative interpretations of gavagai is because from an empirical perspective the four different interpretations of gavagai are on a par. When we are interpreting the verbal behaviour of the native our analytic hypothesis will be framed so as to assume that that they are people like us and hence we will automatically go with the assumption that gavagai means rabbit instead of undetached rabbit part, particular instance of universal rabbithood etc. But this just an assumption and it isn’t according to Quine either justified empirically or something that we can verify empirically. When Quine talks about unverifiable analytic hypotheses he is saying that there is no empirical difference between the reference of the various different translations of the term ‘gavagai’, and in order to begin to translate truth functions, observations etc the field linguist will begin by framing an analytical hypothesis and this hypothesis is not refutable by the empirical data since there is no empirical way to distinguish between the different references of the term ‘gavagai’.

When reading an early draft of Everett’s book I raised the above point with him. In his DMM he replied as follows:

“ As David King (pers. comm.) points out to me, my conclusion here regarding indeterminacy of translation could be open to criticism because it would be possible- and this seems to be what Quine had in mind- for the translator not to know that they had misunderstood, because the native speaker and translator could respond behaviourally in the same way ostensibly, but with different mental maps from experience to meaning-one person responding to rabbit parts and the other to a whole rabbit. In the field, however, this is a difference without a difference. The field researcher does not nor would not notice the problem. But of course it is entirely possible that two people can talk in translatable ways, each one of them buttressed and understanding via distinct dark matters. It is not merely possible but, if I am correct, inescapable.” (ibid p. 341)

In his reply Everett again speaks about ‘rabbits’ and ‘rabbit parts’ but of course as we have seen above Quine’s examples are of a different logical level so Everett’s reply misses to point. Everett’s comment that it is inescapable that two people could talk using different dark matters buttressing them from fully understanding each other is a very Quinian point. Quine would speak of different webs of belief buttressing understanding but this issue is tangential from the ISR. None of Everett’s arguments thus far would have overly troubled Quine.

If we abstract for a moment from Everett’s use of non-Quinian examples to criticise Quine, Everett offers a hypothesis that could conceivably be used as a way of refuting Quine. Everett offers the hypothesis:

“I believe that the absence of these Quinean answers results from the fact that when one person points towards a thing, all people (that I have worked with, at least) assume that what is being asked is the name of the entire object…Objects have a relative salience-whole objects…This is perhaps the result of evolved perception. Perhaps animals perceive wholes before parts. If we are being threatened by a wolf, we are being threatened by the entire wolf, not merely its ears its paws or teeth. And it is likely that a wolf sees a person object when looking at us. We would not last very long in the wild if we saw ears without understanding that ears are part of something else, more important than its parts, that could turn out to be a foe, friend, or food. Initial focus always seems to be on the whole. Perhaps this is to do with biological values of hunger satisfaction, self-preservation, or the like. In any case, it seems to be what happens transculturally. ( ibid p. 266)

This is a perfectly sensible suggestion by Everett, there an enormous amount of evidence that people do indeed make whole object assumptions. Scholars like Carey, Spelke, Markman have constructed many experiments which seem to demonstrate that children have innate concepts of objects and that these concepts will affect how the child will interpret what is said to them. As I discussed in detail in (King 2016a), one difficulty with this argument is that children having innate concepts doesn’t necessarily mean that those concepts cannot change as a result of cultural factors as the child develops. However a critic could argue that Everett’s experiences in transcultural translation show that the whole object assumption is a wide phenomenon. And therefore they could combine Everett’s data with the data from psychologists like Carey to demonstrate that innate constraints on concepts are maintained throughout the life time and occur in all cultures studied demonstrating that Quine’s IDT is false.

But this approach obviously won’t work for a number of reasons. The most obvious reason is because as we have already seen Everett not discovering any natives saying ‘undetached rabbit part’, instead of ‘rabbit’ is to be expected on Quine’s picture; it is no refutation of Quine. But there is another reason that combining Everett’s view with Carey et al won’t work; Everett’s entire book is dedicated to refuting appeals to innate concepts, innate language, innate morality etc. Everett even dedicates a section of his book to criticising the methodology that Carey uses when arguing for innate concepts.

On a superficial level it may seem that Everett is being inconsistent appealing to innate perceptual constants in an attempt to refute Quine while spending the majority of his book criticising appeals to innateness. But Everett is not being inconsistent at all, in fact he spells out his commitments to some form of innate apparatus throughout the book:

“To deny instincts in one domain does not entail denial of the obvious fact that our genes impose strong limitations on us. There obviously are things such as innate characteristics-eye color, adipose cell concentration, blood type, height, and so on. The question is whether there are Bastian like innate conceptual structures…in what follows I want to argue that all forms of innate conceptualism-Platonic a priori knowledge, all Bastianisms- are detriments to understanding, passé and deeply flawed” (Everett: Dark Matter of the Mind pp. 286-287)

“In fact, I believe that there is evidence that humans and other animals may be born with some instincts (candidates include grasping, breathing, making sounds-all non epistemic). Rather, my objection is that the vast majority of research on human instincts is looking in the wrong place. It is looking for knowledge instead of more basic capacities like emotions. (ibid p. 321)

Here we can see that Everett is very clear he has no problem with innate grasping instincts, innate perceptual constants etc, his difficulty is with appeals to innate conceptual abilities or innate apparatus of an epistemic nature. So whether one agrees with Everett’s views or not he cannot be automatically be accused of inconsistency in arguing for innate perceptual constants but not arguing for innate concepts. Nonetheless I will show that Everett is still being somewhat inconsistent as the sceptical arguments he uses to pour cold water on arguments for a language instinct, or a morality instinct are not applied when he argues for the existence of innate perceptual constancies.

It is worth noting that to some degree it is possible for Everett to have his cake and eat it. He could benefit from Carey’s work without being committed to the existence of innate concepts. Tyler Burge in his ‘Origins of Objectivity’ argues convincingly that Susan Carey’s data is evidence of object representations that are perceptual not conceptual. Burge argues that Carey wrongly thinks that because representations as of objects are not reducible to spatial and temporal properties and relations they are not perceptions (‘ Origins of Objectivity’ p. 249). Burge argues convincingly that Carey is incorrect in holding this view and that the fact that our perceptual apparatus attributes bodily representations when objects are out does not show that such representations are not perceptual (ibid p. 249). Whether Burge’s analysis is correct or not is beyond the remit of this paper. My only point is that a theorist who wants to rely on perceptual constants to undermine the IDT may find Burge’s work useful. Everett however is denied this obvious route because he rejects the very evidence that Carey and Burge relies on. Firstly Everett’s just dismisses Carey’s arguments for innate concepts:

“Even Carey’s theory of concepts recognizes that many concepts (e.g. “US president”) must be learned. However, once we have understood how, why, and which concepts are learned transculturally, what is left for Nativism, aside from standard poverty of stimulus arguments? And what, after all, does “poverty of stimulus” mean in practice, other than we cannot think of a stimulus responsible for a particular concept, action or other learning? As many have said in the past, when looked at carefully, the expression “poverty of stimulus” in interchangeable in practice with “poverty of imagination”.” (Everett: ‘Dark Matter of the Mind p. 274)

I have been critical of poverty of stimulus arguments in the past (see King 2016b), but not on the grounds that they amount to nothing more than a poverty of imagination but rather because the primary poverty of stimulus arguments used in linguistics have been refuted. These refutations have typically been either ignored or dismissed as unimportant by generative grammarians. The work on concept acquisition by people like Carey is much more experimentally driven than that of Chomsky,  the constant experimentation that occurs to test claims that x or y is innate, means that there is always a possibility that future experiments will refute a claim that x is innate. While Carey does make a distinction between competence and performance her distinction is not as rigid as Chomsky’s distinction. Carey won’t just dismiss performance data as irrelevant if it contradicts her theory[5]. So there is not much merit in claims that her arguments for innate concepts are mere poverty of imagination, the rigid experimental procedure she uses means that her imagination is constantly kept in check by reality.

Everett though doubts the soundness of Carey’s experimental tests. Everett argues that the Gaze monitoring paradigm favoured by most contemporary psychologists is methodologically flawed. His criticisms of the Gaze monitoring paradigm are little more than a caricature. Everett argues that the method erroneously relies on an assumption that we can perfectly interpret infant’s gazes (DMM p. 308). This claim of Everett’s  is clearly false; a close look at any of the many experiments done in the field shows many controls put in place to avoid bias and shows many experiments ruling out previous interpretations held by the theorist (See Carey 2009 for examples). Nonetheless the upshot is that by rejecting the empirical work achieved in the gaze following paradigm Everett is making it extremely difficult to prove his case that innate perceptual constants rule out the IDT.

As we saw above Everett constructed a quasi evolutionary “just-so” story to demonstrate his case that all humans are born with innate perceptual constants. This is in keeping with the general arguments of his book; a key point he makes over and over again is that if we are to postulate innate apparatus we need an evolutionary story as to how it arrived. In fact Everett offers a list of reasons as to why he is sceptical about claims that certain aspects of human knowledge are innate:

“Here are some of the things that bother me about proposals that important aspects of human knowledge are innate (e.g. Morality, language): (1) The nonlinear relationship of genotype to phenotype; (2) failure to link “instincts” to environment-today’s instincts are often tomorrow’s learning, once we learn more about environmental pressures to acquire certain knowledge; (3) problematic definitions of innatess; (4) failure to rule out learning before proposing an instinct; (5) the unclear content of what is left over for instincts after acquired dark matter is accounted for; and (6) lack of an evolutionary account for the origin of an instinct.” ( DMM p. 284)

Everett’s difficulties with proposals for instincts in the case of language are seriously at odds with his casual postulation of innate perceptual constants. Everett just presented a vague story about the survival value of innate perceptual constants but justified it with no actual data. Contrast this with Chomsky who has co-authored multiple articles and a recent book (co-authored with Bob Berwick) book on the evolution of language ‘Why only Us: Evolution and Language’ going into great detail as to how the purported language instinct evolved. One doesn’t have to agree with Chomsky’s theory ( I don’t), but one wonders why Everett is so stringent in criticising the postulating of a language instinct but doesn’t hold himself to the same standards. On no reading can Everett’s casual “just-so” story be considered a serious attempt to deal with the issue of the IDT. Everett has written serious papers on the evolution of language[6] so he knows what it involves. One wonders why his evolutionary speculations on perception were not given the same seriousness. I suspect that he needed some theory to explain why he never encountered Quine type cases and the “just-so” story seemed to fit the bill. But as we have seen he shouldn’t have been surprised that he never encountered Quine type cases; this is to be expected.

If Everett wants to try and undercut Quine’s IDT using perceptual constants he needs to first present serious evidence for such constants. It is hard to see where this evidence could come from other than engaging with developmental psychology in detail. The caricatures he presents of developmental psychology don’t represent the actual work in the field. More detailed evidence is necessary before he can claim to have refuted Quine’s IDT argument.

 

 

 

 

 

 

 

BIBLIOGRAPHY

Berwick, R., and Chomsky, N. (2016) Why Only Us: Language and Evolution. The MIT Press: Cambridge MA.

Burge, T. (2010) Origins of Objectivity. Oxford University Press: Oxford.

Carey, S. (2009) The Origin of Concepts. Oxford University Press: New York.

Carey, S. (2014) On Learning New Primitives in the Language of Thought: Reply to Rey. Mind and Language 29 (2): 133-136.

Everett, D.L. (2016a) Dark Matter of the Mind: The Culturally Articulated Unconscious. The University of Chicago Press, Ltd., London

Everett, D.L. (2016b) Grammar Came Later: Triality of Patterning and the Gradual Evolution of Language. Journal of Neurolinguistics Nov 10 2016.

Everett, D.L. (2017) Forthcoming How Language Began. New York: W.W. Norton/Liveright.

King, D. (2016a) Indeterminacy of Translation and Innate Concepts: A Critical Review. Linguistic and Philosophical Investigations 16: 104-141.

King, D. (2016b) Poverty of Stimulus Arguments and Behaviourism. Behaviour and Philosophy, 43 36-61.

Markman, E. (1989) Categorisation and Naming in Children: Problems of Induction. The MIT PRESS: Cambridge. MA.

Pinker, S. (1994) The Language Instinct. Penguin: Random House Uk.

Quine, W.V. (1960) Word and Object. MIT PRESS: Cambridge, MA.

[1] Dark Matter of the Mind will henceforth be referred to as DMM.

[2] The Indeterminacy of Translation Argument will henceforth be referred to as IDT.

[3] For a detailed evaluation of uses of the IDT in the cognitive science literature see my ‘The Indeterminacy of Translation and Innate Concepts:  A Critical Review’.

[4] Henceforth the Inscrutability of Reference will be referred to as ISR.

[5] Sue Carey ‘On Learning New Primitives in the Language of Thought: Reply to Rey’, where she correctly criticises Chomsky scholar George Rey of making using competence so rigidly as to rule out any possibility of being refuted.

[6] See Everett’s excellent ‘Grammar Came Later: Triality of Patterning and the Gradual Evolution of Language’, Everett also has a forthcoming book ‘How Language Began’ New York: W.W. Norton/Liverlight.