The Science of Language

by Noam Chomsky

  Page 108, On methodological dualism

  Methodological dualism is the thesis – it's rarely stated explicitly, but it's obvious in the practices of many – that the scientific (naturalistic) study of mind, and especially language, requires a different methodology from the scientific study of anything else, such as the natures of helium atoms. Chomsky sometimes mentions Quine's work when providing examples. Quine held that in the study of language, “behaviorism is necessary.” Language, a crucial aspect of mind, can’t be studied in the way the heart can: you can’t “look inside” or postulate an internal system that allows humans, but not other creatures, to speak. Other examples include Wilfrid Sellars and many of his followers and contemporary connectionist studies of language and other mental capacities. There are many symptoms of the attitude. For example, language is assumed to be a ‘public’ phenomenon, a set of practices within a community, something that people made and continue to remake – an artifact. It cannot then be understood as an organ in the head that grows automatically. It cannot be seen as innate, but must be seen as a social artifact, to be studied as a form of behavior governed by socially instituted training and education procedures. The same applies to the ethical domain: values are taught as ought-to-dos that a social group and its institutions (schools, parents, religious institutions, governments . . .) instill in the young born into that social group.

  Chapter 20

  Page 111, On differences between natural language(s) and natural sciences

  Note that the discussion of the ship of Theseus, of the concept PERSON, of WATER, and the like are within the scope of what Chomsky thinks of as syntax, broadly conceived. They focus on the mind and what it natively provides (its contents and its capacities), not – per impossibile – on the way the world ‘really is’ (or is not) for some science or another. They also quite obviously recall a discussion concerning the limits of the human mind and skepticism that goes back to Descartes and before. A particularly interesting twist in the discussion appears where Chomsky proposes that cognitive science take at least as part of its task the investigation of these limits. As it stands, that is an unusual view of cognitive science.

  Chomsky's point about the difference between natural languages and the symbol systems of formal science and mathematics (and the related point about the difference between commonsense understanding and view of the world compared to those offered by science) should not, I think, be thought of along the lines of the dispute in the philosophy of science about the commensurability of (say) Newtonian mechanics and quantum mechanics. That dispute concerns the extent to which a scientist who understood quantum mechanics could understand Newtonian (not the other way around, surely) and, to my mind, I would be surprised if he or she could not. Like too many philosophical disputes, it is often stated in terms that appear to require deep reflection that invites continued dispute, with no way to decide the matter. The gap between natural languages and the symbol systems of the sciences is another kind of issue; it concerns an empirically decidable question that can be stated in various ways. One of them is: can the child (or adult without learned background in particle physics, or ancient Greek, or Cro-Magnon individual) understand RIVER? Can he or she understand (as opposed to mouthing) HADRON? If the answers are respectively ‘yes’ and ‘no,’ that is evidence of a gap. So are differences in syntax, differences between native system and invented symbol system (an artifact), etc.

  Chapter 21

  Page 115, Externalist study and semantics again

  Chomsky's point here seems to be that of course one can have externalist intellectual projects involving language; studies of language use, of communication, of cooperation, and the like are externalist and – as it turns out – they are in the domain of pragmatics. Whether these studies are or yield sciences is another question: perhaps, perhaps not. On the project of an externalist semantics, as indicated already, Chomsky is clear: so far as we can tell, there is no viable project of that sort. Its practitioners seem to want it to be a contribution to science, but its subject matter is the use of language, and it is not likely that we will ever be able to offer a science of the application of language. One explanation of its failure as such a project lies in lack of recognition of the creative aspect of language use.

  Chapter 22

  Page 116, On externalist study and semantics again

  The banter reflects a serious point that Chomsky has often made, that because we are biological organisms, we have limited cognitive capacities. Philosophers have made the point often by speaking of human finitude. And, generally, they have offered nostrums and worse on how to reconcile oneself to or come to terms with the fact: our lives are tragic but . . .; maybe we can never know, but we can have the certainty of faith (and authority); we’re finite, so we suffer from lack of knowledge, but if we’re good (or at least try very hard), we’ll be rewarded with Nirvana, heaven, seeing the Good, escaping reincarnation, everlasting contemplation . . . Chomsky's apparent view is unusual and refreshing: embrace our limitations; it makes life interesting and challenging.

  Keep in mind that Chomsky sees his work on language as a natural science – one that reveals the remarkable (and undoubtedly useful) capacities of a system with which we alone seem to be endowed. And likely it is not the only one. One candidate is a moral faculty. Another is an aesthetic one.

  Chapter 24

  Page 123, On the mind's operations

  We cannot prove that the mind does not work by deterministic principles. But there is no proof that it does, either.

  There is, however, Descartes's observation: we think we are free to choose alternatives. It is an observation that should be taken seriously, even if it is not proof, by any means. And there is the historical and anthropological evidence that Chomsky mentions below. Along with that, if it is plausible that the mind is made up of multiple systems and action is the result of a massive interaction effect (this is not certain, of course, but it is reasonable, given what we think we know), it is very unlikely that we will ever demonstrate that human action is determined. We have, then, commonsense reasons from observations of our own decision-making and the actions of others to believe that we are free, plus no scientific evidence that we are not, and good reason to think that – because our minds are limited – we will never be able to prove otherwise.

  1 I am grateful to Chomsky for clarifying this point for me in comments on the MS.

  Glossary

  Aitiational semantics

  From Greek aitia: responsible/explanatory factor. As developed by Julius Moravcsik (1975, 1990, 1998) and James Pustejovsky (1995), the basic insight of aitiational semantics can be traced to Aristotle's view of how we view the things of what we take to be ‘our world’ – the world as we experience it. Natural objects – water, trees, animals – are conceived by Aristotle as subject to four ‘causes’ or means of explaining what they are. The explanations/causes are material, efficient, formal, and final. The material explanation says what something is made of, the efficient how it came about (most nearly approximating the usual notion of a cause), the formal its structural description, and the final its end or purpose. For Aristotle, these causes are inherent in objects and our minds ‘abstract’ them from experience. For internalists such as Chomsky, they are due to the ways in which our minds conceive, where these are fixed by the nature of the mind. How they conceive is due to the ways human minds develop or grow. In effect, our commonsense concepts of natural things are native to us, and structure the ways we conceive of ‘our’ world (where this is opposed to the world of science). If something is seen as having a purpose, it is because that is what we take its purpose or function to be.

  Anomalism

  In discussion of the mind, the term “anomalism” is typically understood in the context “anomalism of the mental.” Donald Davidson in his “Mental Events” (1970) claimed that while we can and do think of “physical events” as subject to causal laws with reasonably well-understood “boundary conditions�
� (restrictions on their applications), and while we think of “mental events” as causing and being caused by physical ones, mental events are not seen as subject to causal laws with well-understood boundary conditions. Chomsky has quite a different view of the mental, “the physical,” and the notion of mental causes.

  Biolinguistics

  The current name for the internalist and naturalistic study of language undertaken by Chomsky and others. Earlier descriptions for this methodology include “Cartesian linguistics,” “naturalistic approach” and “methodological monism.” As these names suggest, those who adopt this methodology assume that language is a system in the head that is innate in some sense (grows/develops as other mental systems do) and is to be studied in the same ways as any other natural phenomenon, according to the usual desiderata for naturalistic scientific research. The assumptions appear to be reasonable ones: they and the methodology yield good theories of the language faculty.

  Brain in a vat

  Terminology that developed as a result of a thought experiment: imagine that you are not as you think you are, a person with a body that lives and acts in the world, but are instead a brain in a vat of sustaining fluids that is wired up in such a way that an experimenter feeds your brain the inputs needed to make you believe that you live and act in the world. This thought experiment is a contemporary variation of one of Descartes's: imagine that the mind is controlled by a malin génie (sometimes translated as ‘evil demon’) that deceives a person to an even greater extent, one that could raise doubts about whether 2+2=4.

  Canalization

  C. H. Waddington's term for the fact that phenotypes seem to develop robustly despite variations in various factors that contribute to growth or development. See p. 279.

  C-command

  “Constituent command”: a constraint on the permissible ‘shape’ of a sentential structure. See Appendix VII.

  Compositionality

  Usually appears in the context “compositionality of semantics.” The basic idea is that the meanings of a finite set of words come to be combined by the application of explicitly stated compositional principles (rules, ‘laws’) to yield an unlimited set of sentential meanings. For Chomsky, linguistic semantic compositionality is syntactic and does not involve any relations to things in the world.

  Conceptually necessary

  Intuitively, ‘without which, not conceivable.’ Applied to what one must assume or presuppose in the study of language, it is usually thought that language must be a system that somehow links together sounds and meanings over an infinite range. In theoretical terms, it amounts to the idea that a theory of language must introduce certain “levels of representation,” these being at least (in current Chomskyan understandings of the system) SEM or a semantic interface to “conceptual and intentional” systems, and PHON or a phonological/phonetic interface to perceptual and articulatory systems.

  Condition C

  One of the three conditions in “binding theory,” where binding theory is a statement of the principles that govern whether a pronoun, anaphor (such as herself), or “R-expression” can/must be “bound” by a noun or noun phrase in a sentence. If bound (required to refer to whatever the noun or noun phrase is used to refer to), the anaphor, pronoun, or R-expression can only be used to refer to the same thing as the noun or noun phrase. If not bound, it is “free,” and can or must be used to refer to something or someone else. Condition A says that an anaphor must be bound in the (minimal) domain of a subject, Condition B that pronouns (including she, he, it . . .) are free (although they might be bound), and C that R-expressions must be free. There is no theory-independent way to say what an R-expression is. It is easy to present an example of a violation of Condition C, however. She in the following bad sentence is an R-expression that is construed as bound by Jane: *Shei thinks that Janei is easy to please. See Chomsky (1981, 1986) and Appendix VII.

  Copy theory

  Appears in the context “copy theory of movement.” Chomsky's first efforts to construct theories of language (grammars) introduced what were called “transformational rules” (both obligatory and optional) that “moved” elements of a derived linguistic structure from one position to another in the computation/derivation of a sentence. The terminology of movement (“displacement”) remained through the early stages of the development of the Minimalist Program, where a principle called “Move” was introduced and distinguished from Merge. Chomsky, as the main text indicates, for a long time thought that while Move was a necessary part of a theory of language, it was an anomaly, not a “virtual conceptual necessity.” However, in later minimalist efforts, it is absorbed into (internal) Merge, and is no longer seen as anomalous. Given internal Merge, the copy theory of movement (introduced in Chomsky 1993) falls into place. There is no actual movement of an element; rather, a ‘copy’ (which can be seen as an element that was merged ‘before’ it is re-merged in internal Merge) remains in place. It is ‘seen’ there at the semantic interface, although it is not ‘sounded out’ – it does not appear at PHON.

  Corpuscularism

  As used in this volume, any theory that postulates a set of elements that are taken as primitives for the purposes of whatever combinatory principles the theory deals with. The theory states how these elements can be put together to make complexes. For chemistry, the primitives are atoms, the complexes molecules. For computational linguistics, the primitives are (combinable) lexical items and the complexes are sentences/expressions.

  Distributed morphology

  Any of several versions of ways to conceive of how words might in the course of a derivation be put together out of theoretically defined primitives to yield the complexes we hear or (with sign) see. Generally, those who defend a version of distributed morphology hold that there is no lexicon as conceived in many theories of language, including Chomsky's until at least his (1995b), where a lexicon includes semantic, phonological, and perhaps formal information in the form of lexical “features,” and these features might simply pass through the derivation to appear at an interface (plausible with semantic features), or they are subject to further rules/principles, or they guide the course of derivation. Distributed morphologists hold instead that while one might start with some kind of package of semantic information, the rest of the material placed in the lexicon gets added in the course of a derivation.

  Edge effect

  Generally, effects that arise at contrasting boundaries. It is used in biology and ecology to focus attention on the effects brought about at boundaries or interfaces between items such as forests and surrounding plains. In current linguistic work, it is given specialized uses in phonology. In syntax it could be used to speak of the effects of Merge. If a lexical item (LI) is merged to a syntactic object (SO), the SO becomes the complement of the LI. One could also speak of the effects of Merge at interfaces. At the phonetic interface, internal Merge yields “movement” or “displacement.” (An internal merge takes an LI ‘inside’ an SO and ‘copies’ it to the SO's edge (see Edge feature and Copy theory).) As for the semantic interface, SEM, an internal Merge yields scope and discourse effects such as topic and new/old information. An external Merge yields argument structure and hierarchy. See the main text (pp. 15ff.) and Chomsky (2008).

  Edge feature

  Lexical items (LIs) have properties called “features.” Features characterize the effect of an LI in a computation or at an interface. In a sense, they specify combinatory properties and an LI's “internal” (or “intrinsic”) content – the kind of information it carries, or its character. Most LIs can combine: they combine with other LIs and with derived/generated and checked combinations of LIs that can be called “syntactic objects” (SO)s. Those LIs that can combine can do so because they have some property that ‘says’ that they can. That is their “edge feature.” Why ‘edge’? Assume a derived (generated) SO. When an LI merges with it, one gets this: {LI, SO}. The LI is at the edge of the SO. The result is phrasal structure. The SO becomes what i
s called a “complement” of the LI, and the LI is the “head.” As a result, the head-complement structure of linguistic phrases reduces to a property of Merge.

  Some LIs lack an edge feature. They do not combine with others and are read/interpreted as interjections, such as “ouch.”

  One way to conceive of what edge features of LIs ‘do’ is to think of LIs with this feature as ‘atomizing’ an LI. LIs with this feature are atoms so far as the machinery of Merge is concerned. If LIs are ‘composed’ in some way (perhaps only as members of a set of features) as a result of some kind of combinatory or agglomerative procedure, LI features themselves would be seen as primitives (atoms, corpuscles) of a different order, the ‘elements’ that combine in accord with the principles of LI-formation, whatever they might be.

  Eliminativism

  The position of those who hold that the principles and ‘objects’ of one theory or framework are eliminable in favor of the principles and objects of another. For example, several philosophers have held that chemistry (including biochemistry) is eliminable because it is reducible to the principles and objects of physics. Chomsky often remarks that “accommodation” might be the better term. And he points out that in point of fact, physics in the 1920s and 1930s had to be modified in order to accommodate it to chemistry – not the other way around. Accommodation is motivated by methodological considerations: it is a goal of naturalistic scientific research. Chomsky's theories aim toward accommodation with biology. Eliminativism is typically motivated by a metaphysical goal: the belief that there is just one kind of “stuff.”