Thursday, August 31, 2006

Comments on Ruth Millikan: Language: A Biological Model


Ruth Millikan: Language: A Biological Model Oxford University Press, 2005.

Draft do not quote. This is a draft of a comment to be published in the SWIF Philosophy Review in a book symposium on Millikan's last book edited by Marco Mazzone

Ruth Millikan’s writings have brought a breath of fresh air in the last 20 years of philosophy. One of her central claim is that linguistic conventions should not be modelled on the form of social norms, that is, rules that can be an object of evaluation in a social group. Rather, the normativity of language is of the same kind as the normativity that is found in the biological realm: biological items have each an effect that accounts for their proliferation, and that effect is their function, which they perform more or less effectively. Linguistic items have functions in this biological, nondeontic sense, that is, they have effects that account for their proliferation. Language conventions are stabilized by these repeated effects. Millikan’s “natural conventionality”, as opposed for example to David Lewis’ highly normative model of conventionality, has two simple ingredients: natural conventions are patterns that are (1) reproduced, and these patterns are reproduced because of the (2) weight of the precedent and not because of their superior capacity to perform a certain function in a given situation, which partly explains the arbitrariness of conventions (LBM, p.2). Explaining the conventionality of language in such terms is indeed a major step toward a naturalistic account of meaning and understanding .
The evolutionary framework that Millikan has been proposing since 1984 is also a powerful conceptual tool to articulate biological and cultural evolution in an innovative way, as Dan Sperber and I have argued in a previous paper (Origgi-Sperber 2000). Instead of a memetic view of replication of cultural units through a process that merely parallels that of genetic replication, Millikan’s theory allows for clearer explanation of how a cultural and biological evolution may interact, and how items (linguistic or other) may have simultaneously biological and cultural functions.

Millikan’s notion of reproduction is independent from specific mechanisms, either genic replication or social imitation and is thus suitably general. A linguistic pattern that in particular, such as a word, a verbal mode or an intonation, may be directly copied from an individual to another, inferred on the basis of received instructions (as in education) or reproduced by counterpart reproduction, that is, by fitting in with the pattern another person has initiated. Conventional greetings for example are, she argues, reproduced in this third possible way.
Millikan has introduced a fruitful distinction between two kinds of proper functions, that is, functions whose effects account for the reproduction of an item, direct and derived proper functions. , This distinction, we argued, helps to explain how culture and biology intertwine in the case of language. The direct proper function of an item is the effect that is historically responsible for its reproduction. Sometimes a device, in order to perform its direct proper function, produces items that are adapted to particular contextual circumstances. These contextually adapted items have a derived proper function, that is an effect that contributes not to their own reproduction but to the reproduction of the device that produced them. For instance, language is both a biological and a cultural phenomenon whose evolution is to be understood at a double scale: the biological evolution of our “language faculty” -that is, the ensemble of organs and cognitive dispositions that made language possible as an adaptation- and the cultural/historical evolution of languages. Although Millikan doesn’t develop this point, her distinction can be used in order to understand how a linguistic item may have a direct proper function of a cultural kind, (for example, the function to stabilize a pattern of information exchange between two people) and, at the same time, a derived proper function of a biological kind to contribute to the proliferation of the biological devices that makes linguistic communication possible (Origgi & Sperber 2000).

Although Millikan’s account is a source of inspiration to rethink the articulation of biological and cultural phenomena within a unified framework, she couples her view with an account of linguistic meaning and intentionality that Sperber and I criticized in our paper.
I will call her view here: “extreme externalism”: according to Millikan, language is a way of stabilizing “patterns in truth and satisfaction conditions” (cf. Précis, p. 1) no matter how these patterns are realized by our cognitive processes. Basically, if a linguistic item proliferates it is because the effects of its production are often enough advantageous to the hearer and the speaker. These advangtageous effects consist in coordinating speakers and hearer’s behaviour. This coordinating function corresponds to the conventional meaning of these items. It is thus of little use for Millikan to investigate what happens in people’s mind when they speak or understand insofar as linguistic patterns succeed in stabilizing conformity to use. In particular, it is of no use to the interlocutor themselves. Speaker and hearer may succeed in performing this coordination function through different cognitive processes, As Millikan says in her précis: “The specific psychological processes, ways of recognizing, evidence relied on for application, that support our uses of proper names, of words for kinds, properties and so forth, need not be uniform from person to person for satisfaction conditions to remain uniform”. Positing a number of filters that constrain which linguistic forms will be replicated (such as the anatomy of our auditory system and some Chomskyan grammatical constraints on what aspects of language will be perceived by a child as functionally relevant) is all that is needed at the psychological level to explain how language proliferates and serves its communicative, coordinative, and informational purposes.
Since Grice, the fact that the interpretation of utterances is highly context dependent has been taken as evidence that comprehension consists in inferring the speaker’s meaning (a mental state) and not just in decoding the linguistic meaning. Millikan agrees that interpretation is never just a matter of decoding: the hearer is able to coordinate with the speaker and respond in the conventional way because the context in which the linguistic pattern has been initiated by the speaker allows, in a critical number of cases, the hearer to respond appropriately. However, the hearer doesn’t have to reconstruct what the speaker had in mind, unlike the Gricean pragmatic approach suggests: the environmental conditions in which the communication takes place are usually enough for the hearer to recognise from which reproductive family a particular linguistic token comes from (eg. If a token of “bank” reproduces the word referring to financial institutions, or that referring to the side of rivers) and thus figure out its meaning. For this, in most circumstances, the hearer need not pay any attention to the intentions of the speaker. That is because, for Millikan, linguistic meaning is not in the head: interpreting utterances is just another way of perceiving the world: “interpreting the meaning of what you hear through the medium of speech sounds that impinge on your ears is much like interpreting the meaning of what you see through the medium of light patterns that impinges on your eyes”. So, when communication proceeds normally, the hearer directly perceives the world through the words, and not the speaker’s thoughts and intentions. Roughly, a linguistic stimulus, if it succeeds in performing its function, attunes us directly with a piece of the world, and what happens in the speaker’s mind is of little relevance.
This is indeed a very special view of language. As Millikan acknowledges, that is how most animal codes work (p. 96). A bee dance succeeds in attuning other bees to the location of nectar, as id they had experienced the trajectory themselves. But human languages seem very different from animal codes. They have evolved in social contexts, for a great variety of communicative purposes and not just for transmit information in the narrow sense of the term. Their complex relation with our social cognition is one of the central tenets of the contemporary debate on evolution of language. It is hard to imagine how such a fine-grained social ability would have developed independently of any awareness of the thoughts and intentions of others.
Millikan says that our ability to effortlessly disambiguate most linguistic forms is not due to our capacity of reading other people’s intentions, but to the fact that language attunes us to an external context in which it is immediately perceivable what the linguistic forms refer to. Here is one the examples Millikan discusses in the book, if someone says “Hit-me” while playing blackjack, the environmental context of playing cards is rich enough to allow the hearer to interpret the speaker in the appropriate way, and cause her to give another card to the player instead of beating him, without having to attend to his mental states. I find this example quite puzzling: this is typically an example in which the use of the expression “hit-me” depends on the existence of an explicit, normative convention, of the kind Millikan sees as untypical of conventions in general (cf. LBM, ch.1): the hearer responds to “hit-me” in this circumstance by following a rule of the game that would have not survived without the existence of a social, normative context. But most of the contexts in which we interpret language don’t provide uniquely appropriate responses of this kind.
Nevertheless, given her view of language as a form of direct perception of the world, Millikan insists that, as bees, we are immediately attuned to the environmental cues that it is the purpose of that piece of language to direct our attention to. One of her key arguments on this point is some psychological evidence she refers to in the last part of the book according to which children learn language in that way, that is, by direct perceiving what a word is about without inferring what their parents or instructors have in mind (LBM, 213 and ff). I found this was particularly surprising. My first surprise comes from my own experience with my child Leo, a 5 years old bilingual boy: The only word I can concede he could have learned in a “Millikanian” way, that is, by a sort of behaviouristic training for responding in the appropriate way in the appropriate conditions, is the word “STOP” used to cause him to stop moving forwards in the street . I trained him from the onset to respond to my utterance of the word “STOP” by stopping immediately, and I can use this “command” when I want him to stop while he’s biking or rolling. His reactions to other similar but different stimuli, such as “HALT” or “Arrête-toi” in French would not be as immediate. With “STOP” he doesn’t need to think about what I mean: he reacts to this just by stopping, even in absence of any clear reason to stop (i.e. even if there are no crossroads or traffic-lights). But this is far from being paradigmatic of his way of learning language. It is an interesting curiosity that people may be trained to respond to a few linguistic items as animals might. But it seems to me very odd to define such circumstances as the Normal conditions in which a linguistic item performs its function. But let’s put aside this anecdotical evidence and have a closer look to the psychological evidence Millikan refers to in the book. It is surprising to find Paul Bloom’s work on language learning (cf. Bloom 2000) mentioned at p. 213 among the references in support to her view of language. Bloom is quite explicit in his book (cf. p. 78 and ff) about the role of mindreading abilities in language learning (see also Tomasello 2000). Children don’t grasp what adults say just because they are equipped “with a neuronal organisation that is easily tuned to interpret the kinds of informational patterns that language presents” (LBM, p. 214) and they don’t see through their parents’ words as they would see through binoculars. According to Bloom, children understand words as signs of a communicative intention: without this understanding of the communicative act, language learning would be quite limited.
Also, I am not aware of any evidence that would show that the first experiences of language learning have to do with a sort of perception of the world by proxy of the kind Millikan describes, that is, in which children are told how the world is and experience it through a new medium. Children come to the world within a linguistic community, they listen to language even in their mothers’ wombs and seem very competent in distinguishing between a perceptual stimulus and a linguistic one. They don’t “follow the mental focus of another person” as they would “follow the focus of binoculars or of a camera” (LBM, p. 217). It seems quite plausible, from a psychological and developmental point of view, that they follow the mental focus of that person because it’s a person, whose mind and emotions may be more relevant for their little lives than the world around.

References
Bloom, P. (2000) How Children Learn the Meanings of Words, MIT Press.
Origgi, G.; Sperber, D. (2000) Evolution, Communication and the Proper Function of Language, in P. Carruthers & A. Chamberlain (eds.) Evolution and the Human Mind, Cambridge UP, pp. 140-169.
Tomasello, M. (2000) The Cultural Origins of Human Cognition, Harvard University Press.