by Noam Chomsky
II.1 Common sense and interest-dependent function
In our everyday lives, operating from a very early age within the commonsense framework as we do, we think and speak of the functions of things, systems, institutions, and the like, and assign these things and ‘things’ different functions (often equivalent to jobs, tasks, or roles). It is difficult to insist on a unique function for anything as it is conceived in this understanding of the world. Water as understood on one occasion is something that can be used to drink and slake the thirst; on other occasions, it is seen as something to cool us, to irrigate crops, to wash, to dilute spirits, to swim, to float boats, and so on. A government's function can be seen as introducing laws, but also as satisfying human needs, guaranteeing rights, controlling violence, engaging in defense, and so on. Words and sentences (as understood from within common sense, where these are taken to be artifacts, not natural objects) are used to classify, describe, refer, insist, cajole, make claims, and – a favorite of many, and often claimed to be essential in some way – communicate. A railroad carries freight, transports passengers, provides income to workers, etc. Organic entities and body parts too are seen as having various functions. The skin acts as a barrier to dirt and disease, serves as something to caress or be caressed, to cool the body, to provide for some kinds of communication, and so on. Tigers are objects of admiration, poaching, ecological balance, taxonomy, taxidermy, the sources of potions, etc.
Understood in this way, the functions of things as understood in common sense are functions-for-doing-something or functions-for-serving-an-interest. When we speak of a bird's wings as allowing it to fly, we understand the bird as agent satisfying needs, and when we see them in their display role in a mating dance, we see them as meeting an organism's needs by serving to communicate. When we speak of something as having a function for us in particular, we conceive it in our various efforts to deal with the world and speak about it and the people in it. To assign a function to something is to give it a role in solving some problem or carrying out some task that serves practical (not scientific theoretical) interests. We put the things and ‘things’ of the commonsense framework to various uses in our efforts to serve our interests by using these things to deal with various practical problems, problems often resolved by carrying out actions and procedures – putting out the dog, caring for a child or elderly relative, washing the clothes, describing what happened in a court of law, and so on. Remarkably, our commonsense concepts of these things – our nominal (noun-expressed) concepts of water, governments, skin, railroads, words, etc – seem to invite and even support this kind of flexibility in their use. It is not just that their use is a form of action, and action is free. In fact, the concepts themselves seem to be sensitive to human actions and interests, guiding our thoughts and intuitions about the things that they characterize. Consider the concept WATER, often an object case for Chomsky (1995a, 2000). Water ceases to be water when a teabag is put into it – an absurdity from the point of view of natural science, where there is no water, for the concept WATER is not to be found, only a compound of ‘normal’ hydrogen and oxygen atoms (not isotopes, as with deuterium and tritium, which have isotopes of hydrogen) with some very interesting properties that can only be defined and understood from within the sciences that deal with this compound (see above, p. 156). Neither water nor tea is found in science, but they are in common sense, along with an understanding of water as transformable into tea. That transformation and the substances involved matter to us in our everyday lives; they all serve our interests and actions. And rivers (as the discussion of the main text indicates) cease to be rivers when the water in them is subjected to a phase change that solidifies it and a highway divider line is painted in the middle of what is now solid. Chemical composition is irrelevant: water coming from a tap is water, even though it is filtered at the municipal water works through a tea filter and has exactly the same composition as the tea. That is because what comes from a tap from this source is water, period. It is not clear how concepts such as these can (and do) allow for and in their application are sensitive to the various interests that people have when they conceive of the things in question, employing them in different projects. However, the facts are reasonably clear. We gerrymander the functions for us of the entities and systems they allow us to classify and speak of in terms of our variable (although typically related in some way) projects and tasks and the interests they serve. And commonsense concepts of things appear to allow and support this, at least within limits – limits prescribed, presumably, by the features that make up the concept in question.
In a paragraph above I placed words and sentences among the things of the commonsense framework. That is how they are seen in this domain: they are ‘things’ that come from people's mouths, are written on pieces of paper, and do various jobs for us, serving interests in various ways. We think of them as tools; we “do things with words,” as J. L. Austin (1975) put it. There are important differences between ‘things’ like words and sentences and things like water, tables, and people. Words and sentences, unlike the others, are seen as tools that people use to describe, speculate, convince, classify, question, communicate, and so forth. They are seen as entities that humans use to refer and assert, complain and praise, and – anomalously, assuming that words come ‘out loud’ – silently think and ruminate. They are sometimes used to “speak of” or “refer to” – as is said – things and circumstances in the world. When successfully used in this way, one might say, they represent things and circumstances. They are when so used ‘about’ things and circumstances.
Notice that for this way of putting the matter, words and sentences are not ‘about’ things all by themselves. They have to be used in the relevant ways (successfully, on some accounts) and, in this sense, reference and ‘aboutness’ come about because that is how we happen to use some words and sentences, sometimes. Chomsky, unlike many who do semantics in philosophy and linguistics, takes this point seriously. It is one of the background assumptions of his naturalistic internalist approach to the mind's systems and their operations. His internalism is discussed elsewhere in this volume, as is the point about reference, so I do not pursue these matters here. It is, however, worth mentioning that his picture of language and the mind, a picture painted from the point of view not of common sense but of science, can make sense of why the scientist of mind should take the point seriously, and why if one takes it seriously, we can get some insight into how and why ‘the things of the commonsense world’ come to seem to have such variable functions-for-us. To be brief, assume first that natural languages ‘express’ commonsense concepts; for some discussion, see Chomsky (1975, 1995a, 2000). Part of what is involved in expressing such concepts is making them available to the rest of the mind. To make sense of what that involves, assume (as current science of the mind makes plausible) that the mind is made up of many different modules such as vision (perhaps parts of vision too), audition, and the language system, including its core computational system. Assume further – and again plausibly – that language's computational system combines lexical items (‘words’) consisting of ‘sound’- and ‘meaning’-specifying (or ‘concept’-specifying) features and combines these to make complexes called “expressions” (‘sentences’) that amount to a complex of phonetic features at phonetic/phonological “interfaces” PHONs and a complex of semantic or ‘meaning’ features at ‘meaning’ interfaces SEMs (or LFs). On the ‘other side’ of each interface are other systems that articulate and perceive (with the PHON interface), or ‘interpret’ and ‘understand’ with the SEM one. Focusing on the SEM or ‘meaning’ and conceptual interface and its work, assume that the relevant features ‘communicate’ with and ‘instruct’ various other systems, presumably including vision (connected in some way to object-configuration systems and likely separate visual-configuration systems), plus affective and attitudinal systems, social hierarchy systems, what is sometimes called “imagination,” spatial and temporal locations
systems, and so on. Imagination might offer some resources for a kind of cognitive autonomy – several philosophers in the sixteenth and seventeenth centuries attributed imagination to animals, and said that this afforded them a limited kind of mental creativity – but perhaps the language system alone is capable of going virtually completely ‘off line’ to operate autonomously while still enabling use of the resources provided by other systems – giving humans the ability to speculate and wonder, take their thoughts to any situation at any time, fantasize, engage in all kinds of thought, and so on. We need not assume so, however, to acknowledge that in any case in which a SEM and its conceptual ‘information’ is used (that is, when it plays a partial role in interpretation or understanding), multiple systems come into play and that these can vary from case to case. If so, it is no surprise that the concepts or clusters of semantic information that are expressed at SEM can receive multiple applications, skewed to serve various human interests. So of course commonsense things are seen to have and serve various functions, for they are seen in different ways through the lenses provided by different, multiple systems interacting. Nor is it surprising that one gets a massive interaction effect, with no hope of a determinate, scientific theory of ‘what happens’ on the other side of SEM. Apparently, there is no central, integrating module, a system that does the job that terms such as “mind,” “agent,” and “homunculus” are asked to do.1 Yet there is obviously some form of cooperation: people do succeed at consolidating what they have available to act and to understand one another, at least approximately. So it is likely that the best we can ever do as scientists of the mind is speak of how persons use language and the concepts it expresses to accomplish various things, including referring.2 For some discussion, see McGilvray (2005b), the introduction and text of the 2009 edition of Chomsky (1966/2002/2009), and the discussion of linguistic creativity elsewhere in the current text.
Returning now to the discussion of functions-for-us, some artifact concepts might seem to be exceptions to the idea that there is no single, dominant use or function of a class of entities in terms of which a concept can be specified, and perhaps even defined. Not GOVERNMENT, clearly, but an artifact concept such as CHAIR might seem to specify something like a single essential use, so that even if we put chairs to various uses (to serve various functions related to our interests), the concept CHAIR nevertheless expresses a function that anything that can be understood in any way as a chair must serve. A chair must be something to sit on, we say. It is, after all, made by humans to do this job, we think. So could this be said to be a primary or essential function or role, and CHAIR defined in terms of this function? Aristotle tried something like this with his elements (earth, air, fire, water, ether): they had an essence definable in terms of what they were claimed to ‘do’ in his picture of the universe. Earth falls when released, because that is what it is ‘made’ to do. No one can take this seriously in the sciences now, of course: it presupposes that the things in the worlds described by the sciences are like artifacts. Hadrons would have to be thought to be artifacts not of humans, but of a god. But could it work for some of our artifact concepts – at least those where a maker's intention plays an important role, and there is also a plausible claim to the effect that a single intention is in question? Some have staked quite a lot on this idea: a fair amount of work on the identity of works of art, for example, assigns a heavy role to artist intentions in creating a work. As for chairs, one must grant that “something to sit on” is just one function of a chair among others. We who deal with chairs also conceive and understand them as serving, and use them to serve, all kinds of functions. We stand on chairs, use them to weigh down carpets and to cover their blemishes – or to cover gouges on a floor. We display them as signs of wealth, social status, or preference for a specific style, and so on. Still, although used in many ways to perform many tasks for us, is it not the intention of their makers that they be something to sit on, and if they failed to be something to sit on, they would not be chairs? This too is dubious: for what it is worth, my intuitions tell me a broken chair is a chair. Perhaps more obviously: a chair displayed in a museum with signs indicating that it may not be sat upon is a chair. Moreover, maker's intention with regard to function cannot be all there is to it. Benches, stools, sofas, and loveseats are also made to serve as things on which to sit. Also, one can sit on any number of things, including boulders and branches, and they serve this function, when so used. So there must be something more to defining the commonsense concept CHAIR than appealing to a single function, however ‘primary’ by virtue of the intention of their more utilitarian manufacturers it might appear to be. To return to Aristotle, perhaps at the very least one should speak to the other ‘causes’ of chair: the concept CHAIR should ‘say’ something about their geometry, appropriate materials from which they can be made, and indicate that chairs are (typically?) artifacts. Or consider Aristotle on HOUSE: it is something in which one can dwell (final cause), made of appropriate materials (material), constructed by humans (an artifact: efficient cause), etc. Perhaps adding this other information will satisfy some. But for the scientist of mind and language, it is not clear what the point of the exercise is. It does not succeed as definition as the scientist conceives it. It looks like, and is, an effort to catalogue a rather small group of cases where there is a fair degree of convergence in the ways a population thinks of and uses terms such as “chair.” It is an exercise in what Wittgenstein called “description of use,” rather than a description of the concept as found in the mind, joined to an account of how it developed or grew, and an account of just how it ‘affects’ its use. Perhaps that is what Aristotle was aiming at, rather than describing the use of a term. Perhaps as were later philosopher-scientists such as Ralph Cudworth, he was looking for a way to capture what the idea/concept HOUSE is, and how it ‘works’ (and develops).
Nevertheless, at least some theoretical interests are served by undertaking this and related exercises. For one thing, it highlights the fact that there is a great deal of variation in the application or use of language and that it is to all intents and purposes free, suggesting strongly that use or application of language is not a place to look for success in constructing theories. For another, it offers some data, data that the scientist of language and mind can and probably should take into account. It suggests, for example, that at least in the case of nominal expressions (not adjectives, adverbs, or verbs) and the commonsense concepts expressed in natural languages, one should take into account the fact that our minds tend to look for and expect answers to questions such as “What can it be made of,” “What is it for,” and “Is it an artifact or a natural object?” Julius Moravcsik (1975, 1990, 1998) made very useful contributions in this regard; James Pustejovsky (1995) incorporates some of his insights in a proposal for a theory of language processing. It might also be captured in terms of a view of concepts and their acquisition that ‘breaks them up’ into features – specifically, what Chomsky calls the “semantic features” that go toward specifying the ‘meaning’ side of a person's lexical entry. These matters come up on comments below.
More generally, data from cases such as these and many others got from observing how people use language joined with data from instrumentation of various sorts, language and other system impairment (for example, some of Elizabeth K. Warrington's works [e.g., Warrington & Crutch 2005]), plus neuropsychological and neurolinguistic data, and so on, provide evidence for and against various proposals concerning the architecture, operations, and inputs and outputs of the language faculty and other faculties-modules in the head. Placed in theoretical structures that are making progress, one can begin to understand what is going on in language, vision, object configuration, and so on. I mention these three in particular because there has already been considerable progress with them.
In line with the above, pointing out that there is no single function that language serves also helps undermine some efforts to construct what their proponents call
“theories of language” (in one prevalent form, a ‘theory’ of linguistic meaning), efforts based on the misguided assumption that natural languages have a single use. There are philosophers, linguists, psychologists, and others who hold that language serves the single primary purpose of communication and – even more restrictively – communication of “information.” It is obvious why they want to do this. They conceive of languages not as natural systems, as I-languages or biophysically based systems in the head, but of languages in use – language as it appears in human “practices” and in the linguistic acts and actions of human beings. Looking at language in that way while also trying to construct what aims to be a systematic and unified theory of language, they have to hope that they can find a single canonical use of language. For if they can, they believe, it will display regularities in use and application. If so, perhaps these regularities in use can be made into the rules of language – a particularly tempting prospect to philosophers who would like to use their study of formal logic and inference as the basis for a theory of language. Their preferred strategy makes them look for a functional essence of language, one that they hope to capture by some uniform set of rules (of inference) or conventions that people invent in order to (say) communicate information to each other. Some of their efforts – e.g., those of David Lewis – are discussed below.
