Structural constraints on cataphora

« previous post | next post »

I'm on my way to Pittsburgh for the annual meeting of the Linguistic Society of America. And while I'm waiting for my plane, I think I have just about enough time for a question, even if I fluff it out a bit by giving you the train of thought that led up to it.

Courtesy of The Lousy Linguist ("the germans fear my language too, muahahaha", 1/4/2011), I read Tony Paterson, "'Denglish' now verboten", The Independent 12/30/2010):

Germany's Transport Minister claimed to have struck an important blow for the preservation of the German language yesterday after enforcing a strict ban on the use of all English words and phrases within his ministry.

Peter Ramsauer stopped his staff from using more than 150 English words and expressions that have crept into everyday German shortly after being appointed in late 2009.

What caught my attention in Paterson's article wasn't the silly idea that Germans should be forbidden to use words like handy (meaning "cell phone") and laptop (meaning "laptop"). Rather, it was the final clause in the second sentence ("shortly after being appointed in late 2009"), which I first read as modifying the object of the first clause ("more than 150 English words and expression").

"Wait," I said to myself, "the Germans appoint words? Who did it?" Of course, I quickly retraced my syntactic steps and realized that it's Ramsauer who was appointed, not the words and expressions.

Idly re-arranging the clauses, I was reminded of the anaphoric asymmetry that I first learned about in an undergraduate syntax course many years ago. Roughly, a pronoun in a subordinate clause can precede its antecedent, while a pronoun in a main clause can't:

(a) After [Peter Ramsauer]1 took office as Transport Minister, [he]1 proscribed Denglish.

(b) After [he]1 took office as Transport Minister, [Peter Ramsauer]1 proscribed Denglish.

(c) [Peter Ramsauer]1 proscribed Denglish after [he]1 took office as Transport Minister.

(d) [He]1 proscribed Denglish after [Peter Ramsauer]2 took office as Transport Minister.

There's been a lot theorizing over the decades about how to describe and explain structural constraints on anaphoric relations, which of course are substantially more complex than this simple example suggests. But what I'm wondering now is who first described the basic facts. Did Otto Jespersen write about constraints on cataphora, for example? Or was this a discovery of the first generation of generative grammarians in the 1960s?

I could look it up, but it's almost time for my plane, and an airport waiting room is not the ideal place for such research anyhow. So I'll appeal to the implicit reference library of Language Log readers, several of whom no doubt know the answer.


  1. Mr Punch said,

    January 6, 2011 @ 9:03 am

    Stereotypically, with the Germans, any usage that is not forbidden becomes obligatory.

  2. Murray Smith said,

    January 6, 2011 @ 9:36 am

    In his marvelous little introductory text Language: The Basics, Larry Trask characterizes the general pattern here as "An anaphor may not both precede and command its antecedent." He writes "… nobody succeeded in stating this rule until the American linguist Ronald Langacker worked in out in 1969…" (pages 31 – 31). Unfortunately, he gave no reference.

    [(myl) I dimly (and perhaps falsely) remember learning about this constraint in a syntax class taught by George Lakoff in 1966 or 1967. If so, then either the generalization was around in the oral literature for a couple of years before Langacker's publication; or perhaps some less formally-stated version of it had been around since (say) early-20th-C grammarians like Jespersen. I mention him just because he was a careful observer who studied anaphora along with just about every other aspect of English grammar, and it would be somewhat surprising if he hadn't noticed something as striking as this.]

  3. Faldone said,

    January 6, 2011 @ 9:43 am

    Peter Ramsauer stopped his staff from using more than 150 English words and expressions that have crept into everyday German shortly after being appointed in late 2009.

    Rather than reading this as saying the words and expressions were appointed, I read it as saying that the words and expressions in question had crept into everyday German after Hr. Ramsauer was appointed.

  4. Vance Maverick said,

    January 6, 2011 @ 10:12 am

    I read it as saying that it's OK to use up to 150 English words and expressions, but no more than that.

  5. Mr Fnortner said,

    January 6, 2011 @ 10:14 am

    Without access to the answer key, I would say that (d) is faulty grammar according to myl's "rough rule" (if the utterer meant for he and Peter Ramsauer to be one and the same. If he refers to another, then (d), of course, is a fine sentence. Linguists can supply the requisite labels of cataphora, exophora, etc., at their leisure.)

    My cousin say things like "I spoke to Ned about Dave. He's coming to visit, you know." If I reply, "Who is?" he'll respond, "I told you–Ned|Dave." Either person could be the right answer, and I'm always wrong. What I find interesting is that in my cousin's mind he knows exactly the referent he intends. And I have come to understand that he expects me to implicitly know also. He's a pretty confusing guy to talk to sometimes.

  6. Alexander said,

    January 6, 2011 @ 10:30 am

    Mark, you might be interested in the work on the processing of cataphora, including:

    Kazanina, N., Lau, E., Lieberman, M., Yoshida, M., & Phillips, C. (2007). The Effect of Syntactic Constraints on the Processing of Backwards Anaphora. Journal of Memory and Language, 56, 384–409.

    Kazanina, N. & C. Phillips. (2010) Multiple Routes to Pronoun Resolution in Russian Cataphora. Quarterly Journal of Experimental Psychology, 63.

  7. John Cowan said,

    January 6, 2011 @ 10:43 am

    I wonder about the wider context. Is this a rule of English grammar only, or does it purport to be universal? Does it apply to pro-drop languages when the pronoun is dropped? Does it apply to languages like Chinese, where ellipsized verb arguments are very common? Does it apply to non-configurational languages (assuming you believe in them)?

  8. Spell Me Jeff said,

    January 6, 2011 @ 10:44 am

    I believe there is a close correspondence between my college students' stylistic sophistication in general and their command of cataphora.

    I mean, a student whose style does not extend much beyond SVO or SVO-conj-SVO, seems unable to generate sentences like #2 in myl's examples, and often they are baffled upon reading them. Very often, and despite coaching to the contrary, they will flag as incorrect a structure like the oft-prescribed:

    "In his novel The Sound and the Fury, William Faulkner . . ."

    and are likely to prefer (when reading or writing) the comparable but (to my admittedly jaded ears) more awkward

    "In William Faulkner's novel The Sound and the Fury, he . . ."

    Or perhaps all I'm really saying is that I (and perhaps others) consider the appropriate use of cataphora a sign of stylistic sophistication.

    On the flip side of the coin are VERY basic writers who generate what might be called cataphora in the sense that pronouns precede nouns with no real regard for the sense of what's being communicated. Referents may appear, if they appear at all, many sentences down the road. Composition textbooks refer to the issue simply as "unclear referent" or suchlike.

  9. Luis said,

    January 6, 2011 @ 10:47 am

    @Murray Smith: the reference that Larry Trask did not give is:

    Langacker, Ronald. 1969. On pronominalization and the chain of command. In Modern Studies in English, eds. Schane and Reibel, 160–186. Englewood Cliffs, NJ: Prentice Hall.

    The name that Langacker gave to this pattern is the Backward Anaphora Constraint, which (from memory) states that an anaphor may command or linearly precede its antecedent, but not both. You might also find references to it in Ross' "Guess who?" paper, as it also restricts the distribution of ellipsis sites (qua special cases of anaphora)

    1) Although I don't know who, I know he wants to see someone.
    2) Although I know he wants to see someone, I don't know who.
    3) I know he wants to see someone, but I don't know who.
    4) *I don't know who, although I know he wants to see someone.

  10. Robert S. Porter said,

    January 6, 2011 @ 11:14 am

    I love that they are making a fuss over 'handy' as if it's a genuine Anglicism.

  11. Colin Reid said,

    January 6, 2011 @ 11:53 am

    @Mr Punch: As a result of spelling reforms, there is now an abundance of optional spelling conventions. In many cases, in the first round of reforms a new spelling convention was declared, then in the second round, prompted by public reaction, the old convention was brought back as an optional variant. For instance, the informal second-person pronouns used to be written with a capital letter in correspondence pre-1996, then the uncapitalised versions were preferred until 2006, and now capitalisation is left to the writer's discretion. (At least, this is what I've read.)

    No idea about official verdicts on foreign vocab, though. One thing that impresses me is how quickly German speakers agree on a gender for a borrowed word. How did they decide to make 'Job' masculine, for instance?

  12. John Lawler said,

    January 6, 2011 @ 12:03 pm

    I rather like "backward anaphora" in preference to "cataphora". Use of the latter term is licit among consenting adults, but it shouldn't be used in public without a warning.

    Langacker's succinct formulation of the rule as given by Larry Trask ("an anaphor may not both precede and command its antecedent") has always struck me as a lovely example of the benefits of negative conjunction.

    When I used to give syntax data problems that required students to find and state this rule, very, very few ever came up with this formulation; most went on and on with ors and unlesses until you couldn't figure out whether they were on the right track or not.

  13. YM said,

    January 6, 2011 @ 12:55 pm

    I don't see an ambiguity in parsing, and hence picking the antecedent, if punctuation (or intonation) is available to help:

    Peter Ramsauer stopped his staff from using more than 150 English words and expressions that have crept into everyday German , shortly after being appointed in late 2009.

    (Peter R was appointed)

    Peter Ramsauer stopped his staff from using more than 150 English words and expressions , that have crept into everyday German shortly after being appointed in late 2009.

    (words & expressions were appointed)

  14. Will said,

    January 6, 2011 @ 1:23 pm

    @YM, the comma in the second instance you show there results in a non-grammatical structure. if you replaced "that have crept…" with "which have crept…" it would become grammatical but would also change the meaning — the sentence would then definitely mean what myl originally thought it meant. To punctuate the sentence to remove ambiguity you have to offset the modifier (on both sides) with commas, dashes, or parenthesis. Personally I think dashes would be the most elegant:

    Peter Ramsauer stopped his staff from using more than 150 English words and expressions—which have crept into everyday German—shortly after being appointed in late 2009.

  15. Will said,

    January 6, 2011 @ 1:26 pm

    @YM — I just noticed that your second punctuation was *meant* to change the meaning to what myl originally thought. Nevertheless, I don't think you can use "that" to introduce the clause here.

  16. Adrian Bailey (UK) said,

    January 6, 2011 @ 1:28 pm

    Mr Paterson also writes: "Instead, staff must use their German equivalents: "Klapprechner", "Fahrschein" and "Besprechung" as well as many other common English words that the minister has translated back into German."

    i.e. Staff must use common English words that the minister has translated back into German.

    I guess the editors were on holiday.

  17. YM said,

    January 6, 2011 @ 2:10 pm


    Just to be picky: in my version 1 (with the comma), the relative clause is restrictive. In your version (with the dashes) it is not.

    I personally think 'which' is better, as you have it, but I don't find 'that' ungrammatical here, just a little ugly.

  18. Spell Me Jeff said,

    January 6, 2011 @ 2:17 pm

    FWIW, the hand-held device we yanks call a walkie-talkie was originally called a handie-talkie, and if TV can be believed, the name is still used for similar devices today.

    Conceptually, it is not much of a jump from handie-talkie (or hand-held transceiver and its ilk) to cellphone. I do not know if this was an actual path taken, but it seems plausible, especially since Deutsch and English both use "hand" for hand.

    A good just-so story, at any rate.

  19. Joe said,

    January 6, 2011 @ 2:29 pm

    For anyone who is interested, here is a link to Tanya Reinhart's "Coreference and Bound Anaphora: A Restatement of the Anaphora Questions" (1983). She cites the Langacker article as being published in 1966, by the by.

    [(myl) In those days, it was common for an article to float around in manuscript form (mimeographed or worse) for some years before it was ever formally published. So there may be no contradiction between my memory of having learned about this in 1966 or 1967, and a formal publication date of 1969.]

  20. Mr Fnortner said,

    January 6, 2011 @ 2:43 pm

    FWIW, the participial phrase is troublesome because it is relegated to the end of the sentence. This is better, for those who wish to rewrite, "Shortly after being appointed in late 2009, Peter Ramsauer stopped his staff from using more than 150 English words and expressions that have crept into everyday German." And frankly I think the past perfect–had crept–would be better if one is going to use "stopped." The historical present–stops–could easily take "have crept."

  21. Barbara Partee said,

    January 6, 2011 @ 3:22 pm

    I always thought it was Langacker, but let's ask him and George Lakoff. Between them, they'll surely know.

  22. Neal Goldfarb said,

    January 6, 2011 @ 5:38 pm

    @John Lawler: "Langacker's succinct formulation of the rule as given by Larry Trask ("an anaphor may not both precede and command its antecedent") has always struck me as a lovely example of the benefits of negative conjunction. "

    I don't suppose he'd phrase it that way today. Does Cognitive Grammar recognize the concept of "command"?

  23. Clayton Burns said,

    January 6, 2011 @ 5:54 pm

    An NYT subject for the Linguistic Society of America:

    January 6, 2011, 12:15 pm
    House Reading of Constitution Is Not Without Issues

    "— members of the House of Representatives began to read the United States Constitution aloud from the chamber’s floor…

    …before a word of the preamble could be uttered, Democrats and Republicans battled back and forth over the issue of language, perhaps presaging future partisan battles over the meaning, purpose and application of the document." [End]

    They should not really go at it unless they have linguistic moderation.

    I have been meditating a bit on the rule: "Roughly, a pronoun in a subordinate clause can precede its antecedent, while a pronoun in a main clause can't…".

    "He got angry when people said 'George Bush.' He was trying to go undercover." In a database of 10 billion words, you would probably see sentences breaking the rule.

    Who started it? That person will never admit it.

  24. GeorgeW said,

    January 6, 2011 @ 8:18 pm

    @John Cowan: Regarding the universality of cataphora, I gave some samples (acceptable in English) to my wife, a native speaker of Egyptian Arabic, and she rejected them. She said, "That is broken Arabic, you sound like a foreigner" (A fact she already suspected).

  25. Alexander said,

    January 6, 2011 @ 9:22 pm

    @ClaytonBurns: "He got angry when people said 'George Bush.'"

    This is not a counterexample, since the referent of "he" is not the name 'George Bush'. It is George Bush, who is not referred to in the phrase "when people said 'George Bush'."

  26. J. Goard said,

    January 6, 2011 @ 9:30 pm

    @Neal Goldfarb:

    I don't suppose he'd phrase it that way today. Does Cognitive Grammar recognize the concept of "command"?

    Nope, and the standard reference would be to the work of Karen van Hoek (her 1997 book "Anaphora and Conceptual Structure" from the 1996 paper in Cognitive Linguistics Research).

  27. Clayton Burns said,

    January 6, 2011 @ 9:49 pm

    Alexander, You do not need to have 'George Bush.' George Bush will do. Thanks. Clayton.

  28. baylink said,

    January 6, 2011 @ 11:35 pm

    Handy-talkie is, I think, a Motorola-ism, still common among hams (who generally say HT) and Moto has had many models with that in their model numbers.

    It goes back, I think, to the first low-band HTs in WWII.

  29. Barbara Partee said,

    January 7, 2011 @ 3:12 am

    I got a reply from George: "It was Ron, in 67. I had a preprint of the paper and did it in class. It was as a result of that paper that Haj did his cyclic pronominalization paper, and as a response to Haj I did Pronouns and Reference." So everybody was right about where they first learned it, and the preprint in 67 was the original source.

  30. boynamedsue said,

    January 7, 2011 @ 10:57 am

    I love 'handy', the borrowing of non-existent English words is one of the most interesting phenomena of out global village. Spanish is full of them, and Spaniards get quite irked when told that they can't use 'lifting' and 'a wellness' in English.

  31. copperykeen said,

    January 7, 2011 @ 10:58 am

    I just learned that "anaphora" has a totally different meaning in linguistics than it does in rhetoric. Oh great.

  32. Thom said,

    January 7, 2011 @ 10:41 pm

    @John Cowan. As for Mandarin Chinese, in a conversational sense, it would match similar to English. However, as a caveat, one must be aware of the heavy role of context and word order in Chinese.
    Moreover, the fact that all third-person singular pronouns (he/she/it) in Chinese have the same pronunciation (including tone) demands a higher level of clarity in the word order of the language (as compared to one that depends upon case markings).

  33. Geoff Nunberg said,

    January 8, 2011 @ 2:22 pm

    If memory serves, someone (Jerry Morgan?) pointed out that you can get backwards anaphora across coordinated sentences when there's a causal connection between the clauses, as in "Mary kissed him, and now John is floating on air." The implication is that "command" is a notional rather than syntactic relation.

  34. Suzanne Kemmer said,

    January 8, 2011 @ 3:24 pm

    That is the point also of the work by Karen van Hoek cited by J. Goard (which derives from van Hoek's dissertation work in the early 1990s):
    … the standard reference would be to the work of Karen van Hoek (her 1997 book "Anaphora and Conceptual Structure" from the 1996 paper in Cognitive Linguistics Research).

    C-command and similar structural generalizations are really semantic/pragmatic in nature. Van Hoek collected and typologized all the kinds of violations of the binding conditions noted in the literature, added a large number of others (she had been collecting a corpus of them for years) and constructed a general, single-domain account (framed within Cognitive Grammar) for both the cases that fit c-command or similar structural conditions, and those that don't but whose grammaticality is subject to semantic/pragmatic effects. Structural accounts get as far as they do because the structures themselves are semantic-/pragmatically based — constructions have meaning, as the construction grammarians would say, although assumptions of autonomy of syntax made it undesirable to look for such bases. Anaphora is a discourse phenomenon and is driven by certain kinds of prominence relations between anaphors/cataphors (phors?) and their referents (see Mira Ariel's work on accessibility), even in small-scale discourses like sentences.

  35. J. Goard said,

    January 9, 2011 @ 5:26 am

    @Suzanne Kemmer:

    Thanks for the excellent overview!

    I'm glad that you mentioned Ariel in this context, since her view has at least two significant features that seem to be neglected by much work on nominal reference, but with which I find myself more and more in agreement. In a nutshell:

    (1) Forms of nominal reference are not directly related to external facts about the referents or features of the discourse, but to level of activation in attention and memory systems. It follows that there may be a diversity of external factors explaining a certain distinction (like pronoun versus full nominal) in different circumstances.

    (2) The scale is, for all intents and purposes, continuous. Borderline cases and idiolectal variation are fully predicted.

    I'm hoping that as more and more scholars adopt the view you present, with frequency-sensitive corpus and lab research, we'll be able to gain much deeper insights into anaphora, demonstratives, the definite/indefinite distinction, pro-drop and other phenomena (including in bilingualism and SLA, where I'm focusing) by viewing them together, rather than structuralist approaches that end up treating them separately.

  36. Doug said,

    January 9, 2011 @ 5:28 pm

    We seem to have pinned down the answer to who first stated the rule, but I don't think anyone addressed the larger question of whether anyone mentioned the issue way back in the old days before generative grammar. It would be stunning if no one had noticed or commented on these phenomena before the 1960s, though I suppose that's possible.

  37. J. Goard said,

    January 9, 2011 @ 9:31 pm

    While we're at it, what do you guys think about the Wikipedia entry for "cataphora", which offers

    Finding the right gadget was a real hassle. I finally settled with a digital camera.

    as an example of the phenomenon?

  38. Stephen R. Anderson said,

    January 10, 2011 @ 11:13 am

    @Barbara – a trivial bibliographic note: I have an old paper of Haj's ("A note on 'command'") dated June, 1967, in which he gives "UCSD Mimeograph, 1966" as the date of first circulation of Langacker's formulation. And while George may recall Haj's work as deriving from that, my recollection of the antiphonal syntax classes offered by Haj and George in 1966-67 was that Haj came up with essentially the same idea at the same time (1966) on his own, an example of simultaneous independent invention in science among many others.

RSS feed for comments on this post