Lakoff, George; Mark Johnson;
Philosophy in the Flesh
Basic Books, 1999, 624 pages
ISBN 0465056741, 9780465056743
topics: | cognitive-psychology | language | philosophy | metaphor
A lengthy challenge to the philosophical assumptions underlying the Chomskian position of the supremacy of syntax, leading to universal grammar and everything else.
The opening paragraph clarifies what is being attempted: The mind is inherently embodied. Thought is mostly unconscious. Abstract concepts are largely metaphorical. The first is widely assumed in everyday discourse today. The second is gradually sinking into public consciousness. The third point is trickier; Lakoff and Johnson had already spent a book (Metaphors we live by, 1980) trying to establish it, and most of part 3 of this text is also devoted to this cause. The import of these three points are revealed over the rest of the first chapter, but here's a pithy summary (from an interview by Lakoff): When taken together and considered in detail, these three findings from the science of the mind are inconsistent with central parts of Western philosophy, and require a thorough rethinking of the most popular current approaches, namely, Anglo-American analytic philosophy and postmodernist philosophy." But more particularly, the book highlights how refuting these assumptions invalidates the framework for Transformation grammar (Chomskyan linguistics).
George Lakoff is a prominent linguist, and was initially an enthusiastic and able contributor to the generative grammar movement. By the late 1960s however, he was pushing for expanding the model towards including more semantic notions ("deep structure"), whereas Chomsky felt that the clean lines of his theory would be tarnished by the amorphousness of semantics. Chomsky started with the vision of syntax as autonomous of semantics (or function), but provided a link to semantics via what he initially called "deep structure". At that point, Lakoff, Fillmore and many others felt that this was the gateway to semantics, but Chomsky later repudiated any role for semantics in grammar, thus leading to the bitter "linguistic wars" with the generative semanticists in one side and the generative syntax group under Chomsky and Fodor in the other. Lakoff's main concern here, and also in his earlier books such as Women, fire and dangerous things(1987), is to argue that language is ultimately embodied, primarily by extending physical notions into the realm of the abstract via metaphor. This embodied mind position also denies mind-body duality, and the possibility of "disembodied trascendent reason".
These are three major findings of cognitive science. More than two millennia of a priori philosophical speculation about these aspects of reason are over. Because of these discoveries, philosophy can never be the same again. * Reason is not disembodied, as the tradition has largely held, but arises from the nature of our brains, bodies, and bodily experience.... the very structure of reason itself comes from the details of our embodiment. The same neural and cognitive mechanisms that allow us to perceive and move around also create our conceptual systems and modes of reason. 4 * Reason is evolutionary... not an essence that separates us from other animals; rather, it places us on a continuum with them. * Reason is not completely conscious, but mostly unconscious. * Reason is not purely literal, but largely metaphorical and imaginative. * Reason is not dispassionate, but emotionally engaged.
THERE IS NO CARTESIAN DUALISTIC PERSON, with a mind separate from and independent of the body, sharing exactly the same disembodied transcendent reason with everyone else, and capable of knowing everything about his or her mind simply by self-reflection. ... There exists no KANTIAN RADICALLY AUTONOMOUS PERSON, with absolute freedom and a transcendent reason that correctly dictates what is and isn't moral. What universal aspects of reason there are arise from the commonalities of our bodies and brains and the environments we inhabit. The existence of these universals does not imply that reason transcends the body. Moreover, since conceptual systems vary significantly, reason is not entirely universal. NOT RADICALLY FREE, because the possible human conceptual systems and the possible forms of reason are limited. In addition, once we have learned a conceptual system, it is neurally instantiated in our brains and we are not free to think just anything. Hence, we have no absolute freedom in Kant's sense, no full autonomy. There is no a priori, purely philosophical basis for a universal concept of morality and no transcendent, universal pure reason that could give rise to universal moral laws. NO UTILITARIAN PERSON, for whom rationality is economic rationality... Real human beings are not, for the most part, in conscious control of-- or even consciously aware of-- their reasoning. THE PHENOMENOLOGICAL PERSON, who through phenomenological introspection alone can discover everything there is to know about the mind and the nature of experience, is a fiction. ... we have no direct conscious access to its operation and therefore to most of our thought. Phenomenological reflection, though valuable in revealing the structure of experience, must be supplemented by empirical research into the cognitive unconscious. There is NO POSTSTRUCTURALIST PERSON--no completely decentered subject for whom all meaning is arbitrary, totally relative, and purely historically contingent, unconstrained by body and brain. ... much of a person's conceptual system is either UNIVERSAL OR WIDESPREAD ACROSS LANGUAGES AND CULTURES. Our conceptual systems are not totally relative and not merely a matter of historical contingency, even though a degree of conceptual relativity does exist and even though historical contingency does matter a great deal. The grounding of our conceptual systems in shared embodiment and bodily experience creates a largely centered self, but not a monolithic self. There exists NO FREGEAN PERSON--as posed by analytic philosophy--for whom thought has been extruded from the body. That is, there is no real person whose embodiment plays no role in meaning, whose MEANING IS PURELY OBJECTIVE and defined by the external world, and whose language can fit the external world with no significant role played by mind, brain, or body. ... Because a vast range of our concepts are metaphorical, meaning is not entirely literal and the classical correspondence theory of truth is false. The correspondence theory holds that statements are true or false objectively, depending on how they map directly onto the world--independent of any human understanding of either the statement or the world. On the contrary, truth is mediated by embodied understanding and imagination. That does not mean that truth is purely subjective or that there is no stable truth. Rather, our common embodiment allows for common, stable truths. NO COMPUTATIONAL PERSON, whose mind somehow derives meaning from taking meaningless symbols as input, manipulating them by rule, and giving meaningless symbols as output. Real people have embodied minds whose conceptual systems arise from, are shaped by, and are given meaning through living human bodies. The neural structures of our brains produce conceptual systems and linguistic structures that cannot be adequately accounted for by formal systems that only manipulate symbols. NO CHOMSKYAN PERSON, for whom language is pure syntax, pure form insulated from and independent of all meaning, context, perception, emotion, memory, attention, action, and the dynamic nature of communication. Moreover, human language is not a totally GENETIC INNOVATION. Rather, central aspects of language arise evolutionarily from sensory, motor, and other neural systems that are present in "lower" animals. ... The fact that abstract thought is mostly metaphorical means that answers to philosophical questions have always been, and always will be, mostly metaphorical. In itself, that is neither good nor bad. It is simply a fact about the capacities of the human mind. But it has major consequences for every aspect of philosophy. Metaphorical thought is the principal tool that makes philosophical insight possible and that constrains the forms that philosophy can take. ... we use [the methods of cognitive science and cognitive linguistics] to analyze certain basic concepts that any approach to philosophy must address, such as time, events, causation, the mind, the self, and morality. In Part III, we begin the study of philosophy itself from the perspective of cognitive science. We apply these analytic methods to important moments in the history of philosophy: Greek metaphysics, including the pre-Socratics, Plato, and Aristotle; Descartes's theory of mind and Enlightenment faculty psychology; Kant's moral theory; and analytic philosophy. These methods, we argue, lead to new and deep insights into these great intellectual edifices. They help us understand those philosophies and explain why, despite their fundamental differences, they have each seemed intuitive to many people over the centuries. We also take up issues in contemporary philosophy, linguistics, and the social sciences, in particular, Anglo-American analytic philosophy, Chomskyan linguistics, and the rational-actor model used in economics and foreign policy. 8
Every thought, every decision every act is base upon philosophical assumptions [NOTE: This violates the principle that most thought/action is not conscious] What is real - metaphysics What counts as knowledge - epistemology How the mind works - philosophy of mind Who we are How we should act - ethics 9 [cognitive - AMBIGUITY of the term - some philosophers limit the term to only conceptual or propositional structure - equate it with truth-conditional meaning, external to the body 12 ETYMOLOGY: cog+nomen = with name ] rule of thumb among cognitive scientists : unconscious thought is 95 percent of all thought. 13 the hidden hand of the cognitive unconscious uses metaphor to define our unconscious metaphysics -- the metaphors used not just by ordinary people, but also by philosophers... 14 [unconscious metaphysics: self, time, events, causation, essence, the mind, and morality. ] For the most part, philosophers engaged in making metaphysical claims are choosing from the cognitive unconscious a set of existing metaphors that have a consistent ontology. 14 EMPIRICALLY RESPONSIBLE PHILOSOPHY Historically, philosophy has seen itself as independent of empirical investigation. brought into question by results in Cognitive science - radically new view of how we conceptualize our experience and how we think - consistent with empirical discoveries about the nature of mind. [contrast with "a priori" philosophy p. 22]
faculty psychology: faculty of reason - separate from and independent of what we do with our bodies (perception, motion). [Modularity of Mind thesis] Cartesian / Chomskyan : the autonomous capacity of reason is what makes us essentially human, distinguishing us from all other animals. [This view predates evolutionary theory] Evolutionary view: reason uses and grows out of bodily capacities such as perception / motion. Our sense of what is real [METAPHYSICS] depends crucially on our bodies and sensorimotor processes 17
Even the amoeba categorizes things into food and nonfood... We have evolved to categorize - if we hadn't we wouldn't have survived Categorization is for the most part, not a part of conscious reasoning. 100 bn neurons, 100 trillion connections. Often a dense structure in brain will connect to another dense structure via a sparse set of connections... e.g. eye: 100 mn light-sensing neurons, ==> 1 mn fibers going to the brain. i.e. info in each fiber constitutes a categorization.... A small percentage of our categories have been formed by conscious acts of categorization, but most are formed automatically and unconsciously... PROTOTYPE: Human categories are conceptualized in more than one way, in terms of what are called prototypes. [IDEA: More than one way: Pustejovsky's multiple-inheritance qualia lattice or lcp; Gardenfors' dimensions] Each prototype is a neural structure that permits us to do some sort of inferential or imaginative task relative to a category. - TYPICAL-CASE prototypes are used in drawing inferences about category members in the absence of any special contextual information - IDEAL-CASE prototypes allow us to evaluate category members relative to some conceptual standard. (contrast 'typical husband' with 'ideal husband'). - SALIENT EXEMPLARS are used for making probability judgments Prototype base reasoning constitutes a large part of of our reasoning.
Most categories are also matter of degree (e.g. tall people) - also have concepts characterizing degrees on some scale... with norms of various kinds for extreme cases, normal cases, not quite normal, etc. Such graded norms are called linguistic hedges e.g. very, pretty, kind of, barely, and so on [Women,Fire & DT]. Container metaphor for categories - hides conceptual prototypes graded structures / fuzzy boundaries. EMBODIED CONCEPTS Trivial sense: realized neurally, in a part of our body Strong Sense: concepts are distinguished by their inferential capacity - is a part of (or makes use of) the sensorimotor system of our brains. claims of western philosophy - human reason and concepts are mind-, brain-, and body-free 22 Colour concepts are interactional - arise from interactions of brains, bodies, reflective properties of objects, and electromagnetic radiation. There is in the sky or the grass no blue-ness or green-ness indep of our retinas and brains. 24 sky is blue - but does not even have a surface - colour is not related to "thing-ness" in the world since colours are not things or substances, metaphysical realism fails... subjectivism in its various forms - radical relativism and social constructionism - also fails since colour is created jointly by our biology and the world, not by our culture (which is significant - but as a part of this interaction). 25-6 Colour debate in Philo: Thompson 95; Varela, Thompson Rosch 91
Basic level is usually the middle level of category, not too broad, not too fine. chair - furniture: chair : rocking-chair car - vehicle : car : sports-car Rosch and others mid-70s - middle-level categories are "basic" 1. The basic-level is the highest level at which we have a single mental image - can't get an image of "furniture" but can do for bed or chair. 2. Basic = highest-level at which categ members have similar shapes 3. uses similar motor actions for interacting with categ members [NOTE: does this really hold? 1. - image? or concept? 2. shape? or function? Also doubtful is the claim on p. 30 - link between part-whole structure of a bio genera ==> basic-level categories See also Foundations of Mind critique of "basic-level" categories] Most of our knowledge is organized at the basic level. 29 actions: walking, swimming, grasping social: family, clubs, baseball teams; social actions - arguing emotions: happiness, anger, sadness
row "across" a round pond - 90 deg, ok; 45 deg - maybe?? 15 deg - not! trajector - landmark relations : butterfly in garden [NOTE: LO w.r.t. RO] container schems: "In" - bounded region - e.g. "in the garden" = 3D region boundary, inside and outside - gestalt structure - all three are defined together. Source-Path-Goal schema: trajector moves from source to goal along path Bodily projections: FRONT - see from the front (animals) - move in direction of front (car) - interact with other objects or people (TV or table) relative to us: tree or rock English - facing us; Hausa - facing away from us these relations are not "there" in the world - is part of our bodies. Body usage may be higher in other languages - e.g; Mixtec - no concept of "on": - on the hill ==> located head hill - on the roof of the house ==> located animal-back house [NOTE: Is "animal" a "basic-level" category here?? yet we can image it's back] - i am sitting on the branch of the tree ==> i am sitting arm tree 35
linguistic structures - based on small number of primitive image-schemas: - part-whole - contact - center-periphery - adjacency - link - support - cycle - balance - iteration - near-far - straight-curved - forced motion (push/pull/propel) - orientations: vertical, horizontal, front-back 35 As yet, we do not have any neurophysiological evidence (e.g. PET / fMRI) that mechanisms in perception and movement are also used in abstract reasoning. [NOTE: But should cite imagery results in conceptualization - when we think of an object, the same perceptual areas of the brain light up that are at work when we see it. - Fisk? ] computational models in NTL have been built for * spatial relations - in, on, over, through, under. * bodily movements - grasp, pull, lift, tap, punch * aspectual concepts rel. to structure of actions - starting, stopping, resuming, continuing, and also those indicated grammatically (as in "is running", or "has lifted"). 36
Christopher Johnson: work on metaphor acquisition in children - based on Shem corpus - utterances of a child named Shem [MacWhinney 95] - attempts to discover the age at which Shem used the verb "see" in the metaphorical sense of "know". Prior to using metaphor, Shem goes through a stage in which the knowing and seeing domains are conflated - "Let's see what is in the box" - and later develops into full metaphorical usage such as "I see what you mean" - only when the child is able to distinguish the two conceptual domains. Other metaphorical extensions - e.g. "illuminate" - is learned later based on the primary conceptual metaphor 48 Theory of Conflation: Period of CONFLATION (young children): subjective (non-sensorimotor) experiences are regularly conflated with co-occurring sensorimotor experiences - and for a time children do not distinguish the two. Strong associations are built up. Later, stage of DIFFERENTIATION - when children can separate out the domains, but some of these links remain. Source of "warm smile", "big problem" or "close friend". 46 Grady : Theory of primary metaphor : those that arise naturally through conflation, have minimal structure ==> Complex metaphors form from these primary metaphors via conceptual blending. Universal early experiences lead to universal conflations ==> lead to universal conceptual metaphors. [NOTE: This is not completely culture-independent - note the role of time in the Aymara - recently came to power in Bolivia]
Table of primary metaphors p. 50-54 : (notice "Is" and "Are") * Affection Is warmth: They greeted me warmly [TEMPERATURE (sensorimotor) to AFFECTION (subjective) - based on : Feeling warm while being held affectionately] * Important Is Big: Tomorrow is a big day [maps from SIZE (sensorimotor) to IMPORTANCE (subjective) based on child's experience that big things e.g. parents, are important. [IDEA: this is more likely to reflect a possibly innate part of the perceptual system - bigger objects deserve more attention] * Happy Is up: I am feeling up today [Explanation is weak] * Intimacy Is closeness: We have been close for years, but we are beginning to drift apart. * Bad Is Stinky: This movie stinks * Difficulties Are Burdens: She's weighed down by responsibilities * More Is Up : Prises are high; stocks fell * Categories Are containers: Are tomatoes in the fruit or vegetable group? * Similarity Is Closeness: These colours are close. * Linear Scales Are Paths: John's intelligence is way beyond Bill's. * Organization Is Physical structure: Hod do the pieces of this theory fit together? * Help Is Support: Support your local charities * Time Is motion: after her arrival, the days flew * States Are Locations: I'm close to being depressed and the next thing will send me over the edge. * Change is Motion: My car has gone from bad to worse lately. * Actions Are self-propelled motions: I'm moving right along ono the project. * Purposes Are destinations: He'll make it, but he isn't there yet. ["make it" is mine - what metaphor is it?] * Purposes Are Desired Objects: I grabbed the opportunity. * Causes Are Physical Forces: They pushed the bill through congress. * Relationships are Enclosures: Our relationship is beginning to seem confining. * Control Is Up: Don't worry, I'm on top of it. * Knowing Is Seeing: I see what you mean * Understanding Is Grasping: I just can't grasp transfinite numbers. * Seeing is Touching: She picked my face out of the crowd. 54 Why don't we have metaphors going the other way (e.g. "too much" means "too high")? Narayanan: Because senorimotor domain is more complex - and projections from this complex domain to the relatively simpler subjective domain - are one-way 55 primary metaphors are INEVITABLE - because [Hebbian] learning occurs in the brain based on co-activation 57 [IDEA: Grasp,v. - literal(15) vs non-literal(8) Pull - 55 vs 8 Push - 59 vs 17 See - 617 to 605 (18 senses; top 2 senses 200) Look - 274 vs 165 Pick - 17 (picked her successor +sense 3,4) to 5 (pick flowers/mushrooms+sense5). SENSE 4: "pick a fight/quarrel" - is this semi-idiomatic? CLASS OF metaphorically productive idioms? ]
Complex Metaphors are- built from primary metaphors through conceptual blending - Life is a journey - [The Latin "curriculum vitae" - course of life]: Found a direction in life, not knowing which way to turn - Love is a journey - We have come a long way, but the relationship is not going anywhere. we are at a crossroads - may have to go our separate ways. Complex metaphors are do not have direct experiential grounding - but based on grounded primary metaphors [IDEA: may also be grounded - need to map feature spaces of subjective experience and sensorimotor journey - overlaps - IDEA FOR PROJECT in COG SEM?] Novel metaphors - e.g. "We are driving down the fast lane on the freeway of love" - blend the Relationship Is Containment, Love is a journey ==> Love relationship Is a Vehicle, and lovers are travellers. Fast lane - brings in speed and danger connotations. 66 [Related to the notion of conceptual blending Fauconnier et al]
TWO VIEWS OF MEANING in FIRST-GENERATION Cognitive Science 76 - entirely symbolic - symbolic concepts related to other concepts via formal symbol manipulations - - symbols of thought are internal representations of external reality, objects, properties between them, categories. Both views: MIND is an abstract computer program that can run on any "hardware" - disconnected with the body specifics. THOUGHT is literal, with no room for imagination / metaphor.
conceptual structures arise from sensorimotor experience, giving rise to "motor schemas" and "image schemas". [H. Gardner 1995: The Mind's New Science] 77 Differ vastly in the "a priori" assumptions made by traditional approaches. More than assumption, there is a commitment to make sense of a vast range of phenomena that included polysemy (systematically related linguistic forms), inference, historical change, psychological experiments, poetic extensions of everyday language, gesture, language acquisition, grammar, and iconicity in signed languages. 80 It is all made sense of by conceptual metaphors, image schemas, and radial categories - and by no other theory of concepts yet proposed [IDEA: What about GARDENFORS?]
"Love is a journey" ==> enables many mappings POLYSEMY : words like crossroads, stuck, and dead end - all from travel domain - meanings in the love / purpose domains --? metaphoric POLYSEMY NOVEL CASE : "driving down the fast lane on the freeway of love" DEAD METAPHOR refutation - PSYCHOLINGUISTIC : [Albritton 92] - Love Is A Physical Force metaphor (She knocked me out. I was bowled over. There was a magnetism between us.)  Experiment: Involved literal sentences about love, interspersed with "(3) The attraction between John and Martha was overwhelming. (4) Sparks flew the moment they first saw one another." In priming tests, participants were much quicker in identifying (4) when primed by (3) than by more literal sentences. [NOTE: Could also have been because (3) was more proximal? ] SYNCHRONIC CHANGE - Sweetser: conceptual metaphor provides "routes" for possible changes of word meaning over the course of history. Worked on "seeing is knowing" - data from wide swathe of Indo-European langauges. PROTO-IE root *weid ("see") develops in Greek as eidon, "see" and oida, "know" (root of Eng. idea). In English, becomes both vision words witness and knowledge words wit and wise. [ IDEA: PROJECT: Comparative etymological study - Indian Langs? "we will see ; dekhenge ; dekhchhi ;" ] [DOUBLE METAPHOR: Just as you can't judge Don Bradman by his last innings, it would be a travesty to pass a verdict on Advani on the strength of his final stint at the crease. - Swapan Dasgupta, "Usual Suspects" column, Pioneer, 1/1/06 PUN: Lankans' bad run continues. - Pio 1/1/06] ToI Cal 1/1/06 Maoists caught on wrong foot in maoist minefield Central Teeth for states' Maoist battle ] Unconsciously performed gestures accompanying speech (also thought, see exam halls, designers at work) - trace out images from the source domains of conceptual metaphors. E.g. "can't decide whether to stay home or go out" - gesture - palm up - moving up and down alternately - weighing choices 86 Conflation: child initially does not use see in the "see what i mean" sense - this enters her lexicon at later stage.
p.470 [Chomsky's] early transformational grammar was a reinterpretation of the linguistics from among others: - his teacher, Zellig Harris, from whom he appropriated the idea of syntactic transformations and the idea of headed constructions (what is now called X-bar theory) - Roman Jakobson - the idea of distinctive features (474) and over the years he has incorporated additional ideas from John R. Ross, James McCawley, Paul Postal, George Lakoff, and others with whom he has had fundamental disagreements.
Chomsky's view of language is based on a Cartesian conception of the mind, discussed in Cartesian Linguistics (Chomsky 1966) - key components as adapted by Chomsky: - Separation of Mind and Body - Transcendent autonomous reason: Reason as a capacity of the mind, not the body. Reason as independent of feeling, emotion, imagination, or motor capacity - essences - every kind of thing contains an essence that makes it the kind of thing it is. - rationality defines human nature - there is an universal human nature, an essence shared by all and only by humans. What makes human beings human - the only thing that defines their distinct nature - is their capacity for rational thought and language. - Mathematics as ideal reason - Reason as formal. Ability to reason is the ability to manipulate representations according to formal rules for structuring and relating these mental symbols. Logic is the core of this rational capacity, and Mathematics is the ideal version of thought, because it is the science of pure form. - Thought as language. Descartes (in letter to mersenne) conceptualizes thought metaphorically as language, with complex ideas put together out of simple ones, as sentence are made out of words. Universal reason makes possible a universal language, which would of course have an universal grammar. Language too would be mathematical and therefore purely formal. - Innate ideas. Descartes argued that the mind must have implanted in it by God certain ideas, concepts and formal rules that could not have been acquired via experience (letter to Mersenne, Jul 23, 1641). These a priori structures are just given to us and are possessed by all rational creatures. - The method of introspection. Just by reflecting on our own ideas and the operations of our own minds with care and rigor, we can come to undertand the mind accurately and with absolute certainty. No empirical study is necessary.
Chomsky's "Formalist view of language" - a system of symbols in which individual symbols are individual linguistic elements and well-formed symbol sequences are sentences. Principles for combining symbols or transforming symbol sequences into other symbol sequences constitute the "syntax" of the formal "language".
FORMAL language - developed in logic - Emil Post The symbols of a formal language are meaningless - needs model theory to give meaning. e.g. x = 5; y=7; x+y = 12 - are all meaningless unless we have a model theory that defines =, +, etc. The word "language" in "formal languages" - a metaphorical conception of systems of formal symbol strings - made Chomsky's metaphor appear natural to adherents of formalist philosophy. Indeed, Chomsky took it not as a metaphor for modeling natural language syntax, but as a truth. 475 language must be independent of - memory - attention - perception - motion and gesture - social interaction and culture - contextual knowledge - communication needs of users Syntax, on this Chomskyan account, is the creative part of the human mind. It creates, from nothing external to itself (autonomous), the structures of language upon which rationality is built. Syntax is instantiated in the brain but is causally independent of all nonlinguistic aspects of the brain. The brain is seen as having an "autonomous" syntax module. To be autonomous, it cannot be affected causally by input from any "not purely syntactic" parts of the brain - no inputs that could have a causal effect from any of the above (memory/attention / sensory-motor / cultural etc). [IDEA: possibly the notion of intention is first expressed in syntax and then generates logical form?]
Chomsky views the study of animal communication as irrelevant to any study of the language capacity. [The notion of a separate language capacity (or faculty) is widely opposed among neuroscientists, e.g. see Gerald Edelman's Bright Air,Brilliant Fire: On the Matter of the Mind: there is no neural subnetwork that does not have inputs from other parts of the brain that do very different things. p.209] purely syntactic essence - a set of parameters shared by all languages and known innately by all humans - leaves out many features of most human languages, for example, evidential systems, classifier systems, politeness systems (Brown and Levinson 87), spatial relations systems (Talmy 83), aspectual systems, and lexicalization systems (Talmy 1985b "Lexicalization Patterns"). Working linguists study all aspects of language - meanings of words and constructions, pragmatic, semantic, and discourse constraints on the use of constructions, classifier systems (above list,481)... give the highest priority to the "Distributional Generalization Criterion", that the statement of generalizations is, full descriptive adequacy. (482) Chomsky appears to doubt that all this other stuff can be precisely studied in a scientific manner at all ==> To study something scientifically is to study it using the tools of formalist philosophy - else not rigorous, and therefore, not scientific.
Main-Clause constructions: a class of constructions that occur mostly in main clauses e.g. Here comes the bus (deictic locative) Boy! Is he ever tall! (inverted exclamation) What a fool he is! What idiots we were! (wh-exclamations) who on earth can stop Jordan (rhetorical questions) it's raining, isn't it? It isn't raining, is it? (reversal questions) But Lakoff 87 points out that such "main clause" constructions can appear in final-position subordinate clauses: I'm leaving because here comes my bus. But while both because- and if-clauses take similar ordinary clauses I'm leaving because my bus is coming. I'm leaving if my bus is coming. only because takes the main clause construct * I'm leaving if here comes my bus. The Bulls will win because who on earth can stop Jordan? * The Bulls will win if who on earth can stop Jordan? But no one can stop Jordan would work with both because and if. Also although-, except-, since-, and but- clauses take such main clauses. But not all main-clause constructions go even with because - e.g. imperative or simple question clauses: * you're upset because go home! [imperative syntax; no subject or auxiliary] you're upset because i told you to go home! [indicative syntax] * i'm curious because who stole the money? [interrogative] A generalization to explain this would be of the form: Constructions of type A can occur in subordinate clauses of type B under conditions C. If such a generalization exists, is it a "purely syntactic" one? There is such a generalization. First, the main-clause constructions that can occur in subordinate clauses are those thay convey statement speech acts, either directly or indirectly. All the constructions listed have this property. In fact, even interrogatives (e.g. "who wants to watch a really dull movie?") can be a rhetorical q conveying "no one wants to watch a dull movie". Both the negative and the interrogative can occur in because-clauses, but only when the interrogative construction is a rhetorical question i'm curious because who ever would want to steal the money? (contrast with interrogative form above). But this distinction (type A) - whether a question is a q or a rhetorical statement is a matter of pragmatics! So it is PRAGMATIC generalization that unites the syntactic constructions! And as for the type B clauses in which these constructs can appear as the final adverbial subordinate clause, the clauses introduced by because, although, except, since, and but - are all either reasons for some X, or reasons for not X (e.g. I would stop but who's tired?) So here the generalization is SEMANTIC in nature. Thus, a prima facie syntactic phenomenon, namely which syntactic constructions occur in which final-position adverbial subordinate clauses, is governed by semantic and pragmatic conditions.