Gurus not Saussure about their idol's work

Jacques Guy analyses a half-hearted debunking of linguistic tin gods.

Roy Harris, former professor of general linguistics at Oxford, has published extensively about Ferdinand de Saussure, the Swiss linguist acknowledged by many as the father of modern linguistics. Harris has an axe to grind, and he both grinds and wields it here. In his own words: "It will perhaps seem to some readers that this book might more appropriately have been called Saussure and his Misinterpreters ."

After the opening chapters devoted to Saussure's teachings as reflected in his students' and his editors' interpretations - which were more or less well meaning, but not necessarily faithful - Harris turns to his main targets, in order of increasing power of misinterpretation: Leonard Bloomfield, Louis Hjelmslev, Roman Jakobson, Claude Levi-Strauss, Roland Barthes, Noam Chomsky and Jacques Derrida. From then on the book is not so much about Saussurean linguistics as about the games being played in some academic circles. Are we in for a rollicking good time of shooting down tin gods? No.

First, because the picture painted, of disciplines where incompetence and antipathy reign, is too dismal to invite mirth. Second, because this is too difficult a book to be entertaining, as readers get fair warning in the preface: "My discussion also assumes that the reader is familiar not only with French but with the Francophone tradition of Saussurean studies, including the monumental critical edition of the Cours de Linguistique Générale by Rudolf Engler." This singularly reduces the audience for the book, some might even wryly think, to Harris himself, which is a pity. Readers short of the prerequisites will need patience, relentless concentration and a good command of French, for Harris never provides translations of his many quotations. They will also do well to have at hand a copy of Saussure's Cours de Linguistique Générale ( CLG ) to check the facts reported by Harris, so unbelievably gross are some authors' misrepresentations.

For readers of this review who do not know Saussure's works inside out, let me give a brief overview. First, the CLG is not Saussure's own work. It is the posthumous publication of his students' lecture notes collected and edited by two of them, Charles Bally and Albert Sechehaye. Second, it is remarkably modern - many of the topics tackled amount to problems of computational linguistics, only couched in non-technical terms. But it is best known for the two fundamental distinctions that it brought to light: langue vs parole and signifiant vs signifié .

Saussure hit upon a felicitous analogy for langue and parole : if language be chess, langue is the rules of chess, and parole is the sum of the games played by a community of chess players. Imagine communities where chess is learnt as one does one's native language: by playing and by watching others play. Some players may never witness a piece captured en passant , or they may misinterpret the conditions under which this move is allowed. Either way, they end up playing a different variety of the game, so that there is a continual interaction between the rules of chess as it is played and the chess games being played, each influencing the other. Precisely in the same manner, there is constant interaction between the rules of, say, English ( langue ) and the practice of English ( parole ).

Saussure, however, fell short of providing a good analogy for his distinction between signifiant and signifié . For this, we must turn to Chuang-tzu (or Zhuangzi), a Chinese Taoist philosopher who lived 22 centuries before Saussure. "When you point at the moon with your finger," he wrote, "you do not confuse the moon with your finger. How is it, then, that you confuse things with words when you point at things with words?" The finger pointing at the moon is Saussure's signifiant (often translated as signifier), the moon is the signifié (signified), the two combined - the finger pointing at the moon to mean the moon, not a nearby star - is the Saussurean signe (sign). This is a fundamental distinction, familiar to scientists; no mathematician would confuse pi with a picnic table or a trilithon out of Stonehenge. But it is strangely unfamiliar to many linguists and philosophers: transformational grammar arises from a confusion of signifiant and signifie, and the lucubrations of Jacques Lacan about the square root of minus one have as much to do with its mathematical reality (its signifié ) as pi has with a picnic table.

And indeed, Harris paints the misinterpreters of Saussure as concocting their theories from notions ( signes ) deliberately gutted of their meaning ( signifié ), so that they may insert their own. This is as if Chuang-tzu had written that: "Having pointed at the moon with your navel, you pick the lint out of it, and declare it to be the moon."

Harris does not mince his words either. Of Bloomfield he concludes that either he "manufactured a 'Saussure' whom he could manipulate for his own purposes", aided in this by the knowledge that few Americans would have read the CLG , or that he never understood the first thing about what Saussure was trying to express, and "merely read his own theoretical assumptions into the text". And Bloomfield is, with Hjelmslev, one who receives the gentler treatment. Are Harris's harsh judgements deserved? Regrettably, yes, and the fact is easiest demonstrated by taking one case out of the chapter on Jakobson.

Jakobson had argued that Saussure's principle of linearity was contradicted by the recently discovered notion of distinctive features in phonology. Divested of their jargon, those notions are clear. The principle of linearity is the observation that speech is linear: word follows word, within a word sound follows sound, just like the beads of a necklace are strung in sequence. Distinctive features, on the other hand, are properties of sounds that allow us to distinguish them. For instance, voicing. Try to pronounce "five" and "size" very slowly; voicing is the term for the loud buzzing in your voice box which accompanies the sounds of v and z, but is absent from f and s, and so distinguishes z from s, v from f. Clearly, Jakobson argues, voicing neither precedes nor follows but is simultaneous with the sound uttered, and so Saussure's principle of linearity does not hold. This is preposterous, as the same line of argument leads to the absurd proposition that the beads of a necklace do not form a sequence, because they are distinguished by their individual colour and shape, which do not form a sequence: shape neither precedes nor follows colour. But what is interesting, and rather terrifying, is how Jakobson presents his argument. He quotes a sentence from the CLG that says (my translation): "In speech, words form relationships with one another based on the linear nature of language, which forbids the simultaneous articulation of two elements (page 103)." It should be quite obvious, even in translation, that Saussure used "elements" so as to avoid repeating "words". But even if this were not evident, turning to page 103 of the CLG dispels any doubts, as Saussure briefly discusses there the linearity principle at the level of individual sounds (beads), rather than words or syllables, and draws attention to how stress (colour) fits the model. Jakobson's argument, then, rests entirely on interpreting "elements" literally, out of context, as if Saussure had meant "features", when it is abundantly clear that he meant "words", and on ignoring the invitation to turn to page 103 for further explanation. Coming from a respected linguist, this is shabby, to say the least.

From then on, the misinterpretations get worse and worse, and Harris's forbearance surprises. One expects him to close in for the kill at any moment, but he never does, and he merely presses on with his quest for pearls of misinterpretation, moving from Jakobson to Levi-Strauss, then to Barthes. Even when he quotes Chomsky, who equates Saussure's langue with "the generative grammar internalised by someone who has acquired a language", and who berates Saussure for failing to "conceptualise langue as a generative system", Harris refrains from delivering the easy coup de grâce : on the very first page of his introduction to the CLG , Saussure denounced as prescriptive, divorced from simple observation, and blinkered, the kind of grammar that "only aims at giving rules for distinguishing correct forms from incorrect ones".

Now that is precisely the stated aim of generative grammar: to generate all correct sentences, and only correct sentences. Patent nonsense, since the continual interaction between langue and parole , between the rules of the game and the game as people actually play it, ensures that the set of the rules (grammar) and the set of the games played according to the rules (correct sentences) are forever unstable, forever shifting. Chomsky's and Saussure's views of language are thus fundamentally irreconcilable, and Chomsky's equating langue with generative grammar shows that, like Jakobson with the principle of linearity, Chomsky has not understood the notions of langue and parole , or, perhaps, has chosen not to understand them.

Harris ends with Derrida. Derrida manages to misrepresent Saussure as claiming a natural link between signifiant and signifi é, when in fact Saussure constantly emphasises that the link is purely arbitrary, entirelyartificial. This tour de force is comparable to having Galileo declare that the earth is stationary and at the centre of the solar system. How does Derrida pull it off? Not even through a clever sleight of hand, but merely by taking nine words out of their context and inserting them into his own web of fantasies. One can only marvel at Harris's patience in disentangling the rest of Derrida's logorrhea.

Why Harris chose to write for such a narrow audience in such obscure terms remains a puzzle. Most readers will put this book down in disgust after a few pages, so difficult is it. Harris even seems to go out of his way to make it so. For instance, he warns that Saussure's signifiant and signifi é are quite unlike the medieval notions of signans and signatum , without explaining these recondite terms. You will search for them in vain in David Crystal's Encyclopedia of Language , in the complete Oxford English Dictionary and even more technical lexicons.

Almost every page brings a new instance of such obfuscations. It is as if Harris wanted to make it difficult for readers to contemplate fully, in all its repulsiveness, this obscene image of the "soft sciences" that he has exposed, and so chose to throw a cloak of jargon over its nakedness.

Jacques B. M. Guy is a computer scientist interested in natural language understanding, who holds a PhD in linguistics from the Australian National University.