linguistlist.org>
Editor for this issue: Karen Milligan <karenlinguistlist.org>
ling.ucsc.edu>Despite the trappings of squabble and a charge of "strangely offensive tone", much agreement emerges on matters of fact, if not on matters of taste, from Joybrato Mukherjee's response (LINGUIST 13.1952; henceforth JM2) to the critique by Rodney Huddleston and me (LINGUIST 13.1932; henceforth H&P) of his review (LINGUIST 13.1853) of "The Cambridge Grammar of the English Language." First, a terminological point. The reader of JM2 might be forgiven for thinking Mukherjee's abbreviation policy utterly perverse: he starts out using "CGEL" for The Cambridge Grammar of the English Language (as H&P did), and then switches to using that abbreviation for Quirk et al.'s 1985 work "A Comprehensive Grammar of the English Language". But in fact he is quite right about the very unfortunate coincidence of initials here. To go any further into this matter I am going to have to reveal an insider secret. For ten years The Cambridge Grammar had the working title "The Cambridge Grammar of English", and the abbreviation "CGE". As we were going into production, Cambridge University Press announced we would have to change this, as they felt the title sounded too much like various titles in their EFL series. They wanted it to end in "the English Language" to match their multi-volume "Cambridge History of the English Language." We objected. We believed very strongly that there would be confusion if the initials of our work were the same as those of Quirk et al. (1985). They insisted. So Mukherjee is absolutely correct in noting the unfortunate initials clash (and within his rights, I suppose, to aggravate it by returning to the earlier tradition). There are real grounds for confusion here, just as we predicted. Right now I don't know WHAT abbreviation policy to follow, but clearly I should avoid saying "CGEL" in what follows. On matters of more substance, I discern at least ten points of agreement. (1) BINARITY. JM2 agrees that The Cambridge Grammar does not assume binary branching analyses for everything (Kayne does, we don't). We adopt the division of canonical clauses into subject and predicate (NP and VP in category terms) not because of any infatuation with binary branching but because (like Quirk et al.) we think that it is correct. For VPs like "show me the money" we don't think binary branching is correct; there we assume ternary branching. (2) CONSTITUENCY. JM2 is right to says that we lay somewhat more emphasis than Quirk et al. do on branching (hence on binary branching, where that is the analysis), and constituent structure generally. Our grammar is more firmly founded on phrase structure approaches, and that is indeed a point of contrast between the two grammars. (3) GENERATIVISM. JM2 is correct in seeing much influence of generative grammar on our work. (We did not "resent" his noting this.) We have drawn many insights from generativist work of the last fifty years. We merely pointed out that our adoption of the NP-VP canonical clause structure was not at all motivated by the tenets of generativism. (4) DETERMINACY. We are all in agreement about whether The Cambridge Grammar accepts the possibility of multiple analyses. JM2 accepts that it does, but notes a difference in degree: Quirk et al. tend often to suggest that things are actually indeterminate -- vagueness rather than ambiguity, there being no decision about which is the right analysis in some cases. There is an opposite tendency noticeable in The Cambridge Grammar: we try to find arguments that eliminate indeterminacy and home in on a particular analysis, IF the facts can be found to fully support it. So there is no disagreement here about the approach of the two books. If Mukherjee actually thinks it is proper to remain agnostic about whether "looked after her son" is "[looked after] [her son]" or "[looked] [after her son]", we cannot follow him that far; but we can agree that the Quirk et al. book seems agnostic on the point, and The Cambridge Grammar does not (we give syntactic arguments that the second analysis is the correct one in that case). (5) DATA. There seems to be no dispute about what kinds of data The Cambridge Grammar was based on. We agree with JM2 that it was not based on a selected representative corpus, but rather on large amounts of information about English from various sources, from intuitions to computer-searchable text collections to factual material presented in other publications. We do not quite understand what Mukherjee's definition of a corpus is, but it does not seem to matter here; we can agree with him that The Cambridge Grammar does not appear to be strictly corpus-based in whatever his sense is. Likewise we can agree that we did not attempt to cite sources for every example used. As JM2 puts it, "The reader does not know which of the examples are invented, edited or natural (nor, if they are authentic, where they come from)." Our policy therefore differs from Jespersen's; but (and it is really very surprising that Mukherjee did not note this point) it does not differ from that of Quirk et al.: they, like us, illustrated their points with examples of which many were no doubt real, but none were attributed to sources. (6) STATISTICS. We can also agree with JM2 that "not anything that appears in performance data can/should be included in a grammar", though to say that "there always is a frequency-based threshold level" is going too far: for us, whether something is ungrammatical is still in principle a distinct issue from whether it is rare. Our 1750+ pages of description contain many remarks about frequency as we separate sporadic mistakes from genuinely rare grammatical forms. These remarks are, at least in an informal sense, statistical claims. And JM2 is correct in pointing out that The Cambridge Grammar only very rarely backs them up with exact frequency figures or statistical analyses. Of course, here our policy does agree with Jespersen: he cited literary examples with sources, but did not give statistical frequency data. (7) METHODOLOGY. JM2 states "the authors and the reviewer disagree on two crucial points: the notion of corpus and the theoretical and methodological implications of the use of corpus data." But the disagreement cannot be deep. As noted above, we aren't quite sure what the definition of a corpus in Mukherjee's sense is, so ipso facto we can't really disagree with him about it. And we don't believe we have any real disagreement with him about theory or methodology in corpus linguistics. We appear not to have been doing corpus linguistics in his sense (despite the large amount of work we did, to our great benefit, searching machine-readable bodies of text); we were writing a non-statistically-oriented descriptive reference grammar instead. (8) EXTRAPOSITION. JM2 makes it clear that we are all very largely in agreement about extraposition. H&P and JM2 concur that extraposition constructions are much more common than the corresponding synonymous clausal subject constructions, and that it is nonetheless the clausal subject construction that represents canonical clause structure. JM2 does not appear to suggest that because extraposition is so frequent it must be canonical, and we do not deny that the frequency difference is a significant fact about performance, so there is no real dispute here. (9) VISUALS. On the point about the contrast in number visual aids, we took the liberty of a little ad hominem dig in the ribs against Mukherjee: we noted that as regards his preference for quantitative methods, he talked the talk but (on this issue) failed to walk the walk. We pointed out that a quick count of trees in the two books' coordination chapters suggested that our book had more visual aids than Quirk et al.'s, not less. Mukherjee has now clarified what he meant. He decided to compare the chapters on the verb in Quirk et al. and The Cambridge Grammar, and if not only trees but also all pictorial tablular displays are counted, Quirk et al. come out top. (For the one or two people who will want to actually check this, let me note that JM2's "in section 3.4 no. 2" doesn't mean display [2] in section 3.4 of Chapter 3; there isn't a section 3.4 in our Chapter 3. He means "display [2] in major section 4 of Chapter 3".) With the the details of the competition thus clarified, we cannot disagree with him: Quirk et al. has 29 tables and 15 figures where The Cambridge Grammar has 14 tables and 2 figures in its Chapter 3. In general, our book has fewer impressionistic diagrams of relationships and tends to give more tree diagrams of the structures of expressions in typical chapters (59 separate trees divided between 40 tree displays). Speaking for myself, I have never really liked the arrows and boxes and braces and dots of Quirk et al.'s impressionistic diagrams of meaning. For example, I don't understand Figure 4.51 on their page 221 at all. But some people like these visuals, and Mukherjee is evidently one. That's a matter of taste and cognitive style. What we can all agree on is that The Cambridge Grammar doesn't use such diagrams very much (there are some in Chapter 9, illustrating points about negation, but very few). (10) NEGATIVITY. Finally, JM2 complains that H&P concentrated entirely on the negative points in Mukherjee's review of The Cambridge Grammar. That is correct too. It is in the nature of brief responses to criticism: time and space limitations mean that you gotta accentuate the negative, eliminate the positive, and don't mess with Mr. In-Between. We tackled just a half dozen points where we thought that Mukherjee was making false claims about our book. Those points are now largely cleared up. But Mukherjee's review made numerous statements about The Cambridge Grammar in terms of high and unreserved praise. Some are repeated in JM2. Let me repair an omission by saying that we noted those generous and positive remarks, and I am sure that all the collaborating authors of the book were flattered and gratified to read them. Geoffrey K. Pullum (pullumMail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue