May 20, 2007

Humpty Dumpty linguistics from Robert Fisk

I was surprised to learn from a column in The Independent called "Blair's lies and linguistic manipulations" that Robert Fisk once studied linguistics. Fisk reports:

By great good fortune, I studied linguistics at Lancaster University. Indeed, I read the books of Noam Chomsky, many years before he became a good friend of mine; to be honest, when I read his work, I thought Chomsky was dead.

Eventually Fisk did learn that Chomsky was alive, and met him, and was delighted to find that they shared politics. And in his column dated May 19 he appears to be attempting to connect Chomsky's linguistics to the political views they share. Unfortunately, it rapidly becomes clear that Fisk is totally clueless about the linguistics he claims to have studied at the University of Lancaster while studying for his B.A. degree in English and Classics. It struck me as a little suspicious right at the start that a serious student of linguistics in the mid to late 1960s could possibly have thought that Chomsky (still under 40) was other than highly active, since he was publishing constantly at a staggering rate. And looking at the use Fisk makes of his dimly recollected linguistics classes reveals that he simply was not paying attention — and as usual, did absolutely no fact-checking before he started hammering out angry prose.

The subject of the latest piece of prose hammering is, once again, his hatred and contempt for U.K. prime minister Tony Blair. (New readers should note that Fisk always refers to Blair as "Lord Blair of Kut al-Amara", this being some private joke of his about the proper name to be associated with the peerage that presumably lies in Blair's future; the siege of Kut al-Amara in 1915-1916 has been described as "the most abject capitulation in Britain's military history", so it would not be an honorable name to attach to one's title.) This week's news is that Blair is stepping down, and Fisk says:

Lord Blair is going from us. His self-serving memoirs will, of course, remind us of his God-like view of himself (and, heaven spare me, we share the same publishers) but I doubt if Chomsky's "foregrounded elements" will save him. A "foregrounded element" was something unusual, a phrase placed in such a way that it warned us of a lie to come.

This is not a good account of what a "foregrounded element" is (or was; the odd tense choice seems to be an allusion to the fact that back during his undergraduate days at Lancaster Fisk was taught this definition of foregrounded elements). Those acquainted with linguistics will be surprised to see the word associated with Chomsky's name: foregrounding is a matter of discourse and pragmatics, not of Chomsky's primary subfield of linguistics, syntax. I for one cannot recollect him using the term "foregrounded element" ever, not even a single time. Chomsky is famously uninterested in pragmatics, stylistics, or the analysis of discourse. I recall being told by someone who heard him lecture at the University of Colorado at Boulder that he dismissed the analysis of videotaped conversational discourse as a waste of time, not fit to be included in the language sciences. He has often mentioned in interviews that he does not connect his linguistics to his politics — he does not think there is any intellectually serious way in which linguistics can contribute to the analysis of political propaganda, for example.

"Foregrounded element" is a familiar term for those who study discourse rather than individual sentences, style rather than grammar, and information structure rather than syntactic structure. The primary use is in describing what happens in cleft sentences (see The Cambridge Grammar, pages 1414-1416, in the chapter headed "Information packaging"). Compare the most neutral way of saying that Oswald assassinated Kennedy, as in [1a], with the two cleft sentence versions in [1b] and [1c], in which the foregrounded element is underlined:

[1] a. Oswald assassinated Kennedy.
  b. It was Oswald that assassinated Kennedy.
  c. It was Kennedy that Oswald assassinated.

While [1a] would be suitable for use in a context where nothing was presupposed (for example, as the answer to "What happened next?"), [1b] would be suitable for a context in which someone who knew about Kennedy's death but had the wrong idea about who the assassin was, and [1c] would be suitable for a context in which someone who knew about Lee Harvey Oswald's crime but had forgotten which president had been his victim. As The Cambridge Grammar points out (page 1416), the part of the sentence following that gives a presupposed part of the meaning in the form of what logicians call an open sentence (a sentence with a variable substituted for one of its noun phrases), and the noun phrase preceding that provides the most important (hence foregrounded) contribution, the correct value for the variable. Example [1b] presupposes "x assassinated Kennedy", and contributes the crucial information that x = Oswald.

It is possible that Fisk may have confused foregrounding with fronting. Fronting is a syntactic matter: positioning a constituent of a clause at the beginning rather than where its grammatical function might have suggested it would go. The underlined elements in these sentences are fronted:

[2] a. Her, he tends to ignore.
  b. Who do you love?
  c. To my son James I leave the balance of my estate.

Sometimes fronting a phrase syntactically is a strategy for foregrounding it in terms of information structure, but the two should not be equated. In [2a], for instance, I am not presupposing that he tends to ignore some person x and asserting that x = her; almost the opposite is going on. It is presupposed that there is some female person x, and the new contribution to the discourse — left till last to make it salient — is that he tends to ignore x. Foregrounding her (as in It is her that he tends to ignore) has an entirely different pragmatic effect.

Emphatic stress can (in speech) also accomplish foregrounding: to say Oswald assassinated KENNEDY, with heavy stress on that last word, has an effect somewhat similar to using the cleft version It was KENNEDY that Oswald assassinated. So we have to take it into consideration that Fisk might have meant to refer to emphasis.

Fisk's own characterization of what he meant by a foregrounded element, "something unusual, a phrase placed in such a way that it warned us of a lie to come", is somewhere between the weak and the hopeless; and his suggestion that Chomsky has something to do with the definition of the term is just incorrect. But the worst thing is that every single one of the examples he gives makes it clear that he doesn't know what he's talking about.

His first example is bafflingly irrelevant: it is George Tenet's gloss on what he was referring to as a "slam dunk". He just slings this in as a warm-up. You can read it for yourself in context, but Fisk doesn't even try to connect it up to foregrounding, and nor will I.

The first claim Fisk makes that does relate to foregrounding concerns a Beirut newspaper which

quoted our dear Prime Minister as saying that he was very angry that a review committee had prevented him from deporting two Algerians home because their government represented a "different political system". The "foregrounded" element, of course, is the word "different". This is the word that contains the lie.

I note that he has suddenly departed from his own definition: the foregrounded element was supposed to have "warned us of a lie to come", but now it "contains the lie". That is very different. But never mind. The main thing is that different is not foregrounded in Fisk's sense or in any sense at all. It is an adjective functioning as an attributive modifier, and it is in exactly the expected position, not placed in some special way. It is not foregrounded or fronted or highlighted or emphasized or anything else that I can imagine he might have intended. What he wants to say is that he is furious that Blair should just say "different political system" when what Algeria actually has is (according to Fisk) a political system that "allows it to torture to death its prisoners." He raves for a paragraph or two about how disgusting the treatment of prisoners is, and gives horrible details about how prisoners are tortured or raped to death; and all that may well be just as he says it is. But in that case different is an understatement or a piece of evasiveness rather than a lie, and — more relevant here — foregrounding has nothing to do with what he is talking about.

Further examples of "foregrounding" now come thick and fast, and now, bafflingly, foregrounded elements are called "Chomsky foregrounded elements":

Putting the country first didn't mean "doing the right thing according to conventional wisdom" (Chomsky foregrounded element: conventional) or the "prevailing consensus: (Chomsky foregrounded element: prevailing). It meant "what you genuinely believe to be right" (Chomsky foregrounded element: genuinely). Lord Blair of Kut al-Amara wanted to stand "shoulder to shoulder" with Britain's oldest ally, which he assumed to be the United States. (It is actually Portugal, but no matter.) "I did so out of belief," he told us. Foregrounded element: belief.

Am I alone in being repulsed by this? "Politics may be the art of the possible (foregrounded element: may) but, at least in life, give the impossible a go." What does this mean? Is Blair adopting sainthood as a means to an end?

Conventional: adjective in the ordinary position for serving as attributive modifier, not in any sense foregrounded.

Prevailing: adjective in the ordinary position for serving as attributive modifier, not in any sense foregrounded.

Genuinely: adverb in the ordinary position for modifying a verb phrase, not in any sense foregrounded.

Belief: noun functioning as head of a noun phrase in the ordinary position for the object of a preposition (following of), not in any sense foregrounded. (It is the last word in the quoted sentence, incidentally, which makes it baffling how it could be an element that "warned us of a lie to come", which is what Fisk's definition says.

May: modal auxiliary verb functioning as head of a verb phrase in the ordinary position to be predicate of a tensed clause, following its subject, the noun phrase politics; may is not in any sense foregrounded.

Not a single one of the words he singles out as "foregrounded elements" are foregrounded in any sense that someone who does syntax or semantics or pragmatics would recognize. You could do an empirical experiment on this: type out the sentences involved and take them to some linguist who has not seen this post and ask them to underline one word in each to indicate the foregrounded element. You don't even need to do the typing, I'll do it. Here they are; just copy and paste:

  1. He was very angry that a review committee had prevented him from deporting two Algerians home because their government represented a different political system.

  2. Putting the country first doesn't mean doing the right thing according to conventional wisdom.

  3. Putting the country first doesn't mean doing the right thing according to the prevailing consensus.

  4. It means doing what you genuinely believe to be right.

  5. I did so out of belief.

  6. Politics may be the art of the possible, but, at least in life, give the impossible a go.

The results of this forced choice test will be, I predict, roughly the same as if you chose nouns, verbs, adjectives, and adverbs using a pigeon or a monkey or a random number generator. Fisk may know a lot about the Arab world and the terrible things that have been going on there, and his hatred of Tony Blair may be rooted in real political arguments, but in this column he is just frothing. It is clear that he did not pay attention to his linguistics instructors at the University of Lancaster back in the 1960s, and he didn't look up any of the old notes he took as an undergraduate. He wrote his piece using "foregrounded element" to mean just what he wanted it to mean: something like "word which made me feel when I read it that I was looking at a disingenuous claim", or something along those lines. His grasp of syntax and information structure is nonexistent, and his semantics is that of Lewis Carroll's Humpty Dumpty. The linguists in Lancaster's eminently reputable Department of Linguistics and English Language must be a bit embarrassed that he mentioned them.

Update: The date of Fisk's degree from Lancaster appears to have been 1967; a piece from Selves and Others, July 2nd, 2005, which Mark Liberman located, says: "I belong to that generation of undergraduates who cut their teeth on linguistics. Lancaster University in its second year of existence", says Fisk; "Class of '67, if I'm not mistaken." In that piece, Fisk says this of Noam Chomsky:

Less famous then than now, he it was who introduced me to the "foregrounded element". "Foregrounded" is when someone places words in such an order that a new meaning is attached to them or deliberately leaves out a word that we might expect. The big bad man emphasises the meanness of the man. But the bad big man makes us think of size. "Big" has been "foregrounded". Real linguists won't like the above definition but journalists, I fear, sometimes have to distort in order to make plain.

Foregrounded, as understood in modern linguistics, means nothing like what he says here (which in any case is not the same as what he says in the Independent piece quoted above); and Chomsky (I am prepared to offer a modest bet) never said anything like this about foregrounding in the whole of his life.

Much more likely (as Mark points out to me) that Fisk got the term from a source presenting ideas something like those of (for example) Gerald Bruns. In his 1974 book Modern Poetry and the Idea of Language (too late to be the right source, but suggestive of what might have gone before), we find the following passage, quoted here, which has the right kind of muddle to be the sort of thing that could have inspired Fisk, and even a link (thoroughly inappropriate, in my view) to the early work of Chomsky:

The implication here, of course, is that in poetry the aesthetic experience is finally an experience of language itself. It is this idea of poetry which was taken up and developed on a systematic basis by the Prague Structuralists, who extended the traditional theory of linguistic functions or purposes (referential, conative, emotive) so as to include those utterances in which language is used intransitively. In place of Shklovsky’s ambiguous distinction between prose utterances and poetic speech (that is, between "tortured" and "easy" discourse), the Prague Structuralists, particularly Bohuslav Havránek and Jan Mukarovsky´, formulated a distinction between those utterances in which language is "automatized" according to the economy of everyday speech, and those in which language is "foregrounded." Foregrounding, according to Havránek, is "the use of the devices of the language in such a way that this use itself attracts attention and is perceived as uncommon, as deprived of automatization, as deautomatized, such as a live poetic metaphor (As opposed to a lexicalized one, which is automatized)." Thus, for example, Noam Chomsky’s happy line, "Colorless green ideas sleep furiously," is a foregrounded utterance. Although, as it happens, the sentence is perfectly grammatical—Chomsky composed it to show that meaning is not a necessary effect of "grammaticalness"—it is afflicted or, like many lines of poetry, blessed with a dissonance between lexicon and syntax that renders it impervious to whatever effort we may make to impose an interpretation upon it. The structure of words by which Chomsky’s utterance is constituted occupies, that is to say, that "foreground" of the utterance that is ordinarily the special domain of meaning.

Nearly all the ingredients are there: Fisk cites an example illustrating the salience an attributive adjective gains when it is positioned out of the usual order; Bruns connects foregrounding in the literary sense of "deautomatization" (that is, "use of the devices of the language in such a way that this use itself attracts attention" — utterly different from any of Fisk's efforts at defining the word) to a Chomsky example sentence (used by Chomsky to point out that grammaticalness and meaningfulness are distinct, not to illustrate foregrounding in any sense), and it has a couple of attributive adjectives in it... Mix all this together carelessly in a bowl and add a third of a bottle of whisky, and it should be possible to get something like Fisk's chaotic state of understanding.

But I am just speculating, of course; whether Fisk ever saw any source containing stuff like the above ideas of Bruns is not known (though it would appear that Bruns has been writing on similar topics since the 1960s). And I'm not sure I would trust a man who can't definitely remember the date of his own B.A. ("Class of '67, if I'm not mistaken") to give us a self-report.

Posted by Geoffrey K. Pullum at May 20, 2007 04:17 PM