In JGL 18.4 a review of my book (te Velde 2005) appeared, written by Timothy Osborne. In this Discussion Note, I would like to respond to this review, addressing primarily section 4 (Discussion).

First of all, I want to acknowledge how difficult and potentially hazardous the task of reviewing a book of any sort can be. A reviewer deserves recognition far beyond the free book received. Given the difficulty and potential pitfalls of the task, however, it seems to me that a reviewer should be particularly careful when delving into major criticisms of a book under review. I do not believe this kind of care was taken with this review; the result is a misrepresentation of my work, and for that reason I am responding to it here.

Before addressing the problems I have found, I would like to mention several strong points in Osborne's review. First, his summary of the innovations of my theory and how they fit into the Minimalist framework is generally good. Second, he understands the main issues and presents relevant examples for the most part. Third, his knowledge of coordinate structures and the challenges they present is clearly demonstrated, although his terminology at times diverges from the “standard” terminology used in the framework that I assume. This might cause some confusion for readers who expect the predictable, time-honored terms. One example is his sentence: “the approach strives to account for the structure and behavior of coordination via the principles and procedures that are already acknowledged for non-coordinate structures.” Although the central point of this sentence is probably clear, it is misleading to state that my proposal assumes coordination has a “behavior” and that the derivation has a “procedure” of some sort; more appropriate terms would have been “properties” and “derivations,” respectively. This inappropriate terminology is just the first indication that Osborne is not approaching my book from a generative, Minimalist perspective. While it may not be a problem for some reviewers to evaluate another person's work from the vantage point of a very different framework, I believe this is not the case with Osborne's review. As I argue below, this bias results in a misreading of my account of coordinate ellipsis in chapter 4.

The first critical comment in Osborne's review concerns the amount of redundancy in my book. He states that “there are […] a number of significant problems with te Velde's theory and the manner in which it is presented in the book. The presentation suffers from redundancy.” As I point out below, the redundancy that he finds did not prevent Osborne from overlooking some rather obvious points of my proposal. This comment also contrasts starkly with one made in Hagerty's (2006) review of my book where he states: “the approach [of te Velde's book] is developed in layers, with a brief survey in chapter one, an extensive overview in chapter two, and detailed exposition in chapters three and four. There is not much repetition since the granularity of the exposition changes significantly with each level” [emphasis mine]. So whose assessment is right, Hagerty's or Osborne's? This is quite a subjective matter, since every reader brings different background knowledge to the reading. Osborne's opinion stems from the fact that he had spent several years dealing with issues of coordinate structures before reading my book, while Hagerty had devoted comparatively little time to them. It is quite clear from the outset, I believe, that I wrote my book for a person with little or no knowledge of the issues. Osborne should have kept this in mind when writing his review, along with his own vantage point of having studied the issues in a different theoretical framework for years before writing his review.

Osborne continues with a criticism of the organization of the book, namely, the fact that 41 pages of notes follow the last chapter. He states: “working through the book hence requires a lot of flipping back and forth.” To whom should this criticism be directed? My original manuscript had the notes as footnotes, and they would have stayed in this position, if the choice had been up to me, for Osborne is right, the placement is not good, but this criticism should be directed to the publisher. Osborne should recognize that authors do not always have final say in such details.

More substantive problems with Osborne's review have to do with his comments on my data and how I propose to account for them. Osborne claims that the grammaticality judgments on some of the constructions are “debatable.” He discusses, for instance, the construction in 1.

He points out that Ross (1970, 1976) and Sag (1976:145) find such constructions marginal, and in a footnote he discusses two possible interpretations of this construction. He also points out that I assume only the one interpretation in my analysis. However, his discussion of this construction misses the point entirely because the construction occurs not in a discussion of Gapping (chapter 4), but on page 3 of the book where it is used to give a simple illustration of how matching in coordinate structures with ellipsis can facilitate the interpretation indicated by the analysis. My point on page 3 is that through matching, one interpretation can be eliminated, and thereby the ambiguity as well.

Osborne also questions grammaticality judgments on constructions that are clearly taken from other studies—the references immediately precede them, for example, iia,b in note 54 on page 344. Although the use of data and the concomitant judgments constitutes an endorsement of the judgments, criticism of the judgments should acknowledge that they are not mine alone, but ones that are widely accepted in the literature. Osborne does not do this.

The issue of grammaticality judgments on elliptical coordinate structures is a very thorny one; this should have been pointed out in the review. The reasons that reliable grammaticality judgments are more difficult in the case of these structures stem from two facts. One is the symmetry requirement on the interpretation that I investigate in detail in my book. This requirement occurs on the perceptual side of language processing, thus the listener or reader must recognize certain coordinate symmetries and the role they play in the interpretation of elliptical coordinate structures. Without this recognition, the elliptical construction cannot be processed or interpreted grammatically. Furthermore, such coordinate symmetry is not determined just syntactically, semantically, or pragmatically; it results from a combination of syntactic, semantic, and pragmatic factors. In 2a, for instance, an informant making grammaticality judgments must be sensitive first and foremost to the parallelism between the phonologically spelled-out finite verb in the first conjunct and the gap in the second conjunct. In 2b, there are two gaps and two antecedents, further complicating the symmetry.

In addition, there must be a phonological component in this construction for the interpretation of 2b as indicated in this analysis, namely, there must be rising intonation at the end of the first conjunct, and there must be a pause where the two gaps are indicated. Without these two prosodic features, 2b can be interpreted differently, with Joe as an indirect object, in which case there is no ellipsis (in my analysis). But this interpretation also relies on a parallelism in the prosody for complete clarity: both indirect objects need to be stressed.

Because of the role of prosody in the interpretation of the data, such constructions are ideally elicited verbally, not in written form. If verbal elicitation is not possible, one of two problems arises: If the sentence is judged grammatical, it is not certain what interpretation the informant assumed; if the sentence is ruled ungrammatical, it is not clear what caused the ungrammatical rating. These problems are even more pronounced in dealing with constructions that have one or more embedded clauses.

Another issue concerns the value of constructions that are marked as less than fully grammatical (or are “debatable” in Osborne's words). In my book I did not use any grammaticality marker to indicate “marginality,” other than the standard “?” and “??.” A very interesting study by Featherstone (2005) has shown quite convincingly that a more refined system of grammaticality judgments based on magnitude estimation can yield important insights into the grammar of a language under investigation. The point here is that marginal constructions can have as much worth to an empirical study as the grammatical ones.

A final issue related to grammaticality judgments is the bias of the investigator. Admittedly there were instances when it might have been better to use the markers “?” and “??,” but did not simply because the constructions seemed to me—and at least to some of my informants—to be perfectly grammatical, whereas they might have been judged by other informants as odd or marginal. The reason for this disparity can be found in the first point discussed: an informant not familiar with the properties of coordinate structures would tend to rate such structures as less grammatical than one who is sensitive to the role of parallelisms in the interpretation of coordinate structures with ellipsis. For this reason, I chose informants who were usually familiar with the construction types and were often linguists themselves and understood the issues; this may or may not have always led to the best judgments, depending on one's theory on judgments. In short, the greater difficulty with eliciting grammaticality judgments on elliptical coordinate structures is a problem that Osborne should have acknowledged, if he was aware of it. If he was not aware of it, the incisiveness and fairness of his review is diminished.

Let us move on to the main criticisms. Osborne evaluates my account of ellipsis in clausal conjuncts. My assumptions about what types of ellipsis occur follow very closely standard assumptions in the more recent generative literature, a point that Osborne did not realize or did not acknowledge. For instance, his discussion of example 3 does not mention that this construction comes from Schwarz's (1998) study, who discusses coordinate structures first introduced by Höhle (1983, 1990), which are in turn analyzed by Johnson (2002), who focuses on the same issue that I address.

Osborne misses the main points of my analysis entirely. First, I account for the contrast between 3 and the ungrammatical 4, and second, Johnson does not account for this contrast, due to the negation scope of keiner. Because this negation scope extends into the second conjunct, the interpretation of er is rendered illogical; only keiner can be logically interpreted as the subject of the second clause (and thus of both clauses). In short, the ungrammaticality is not due to the structure per se, contra Johnson.

Furthermore, Osborne states incorrectly regarding 5 that “the pre-deletion reading […] involves two people, whereas the post-deletion reading involves just one” (p. 331).

Discussion of the subject gap in constructions like 3 and in related ones like 5 and 6 occurs several times in the book. The point is consistently made that a subject gap needs to be posited in 5, based on certain constructions in which disjoint reference between the lexical and the gapped subject is required for interpretation.

Osborne states that “te Velde thus emphasizes the significance of the exception in the debate, and in so doing, deemphasizes the rule” (p. 333). I believe I make it abundantly clear that constructions like 6, in which the disjoint reference reading is required, are atypical. Furthermore, calling 6 “the exception” misrepresents a core element of my proposal, namely, that symmetry in elliptical coordinate structures is crucial to interpretation; if the listener does not interpret a subject gap in these constructions, a key element of the meaning will be missing, specifically, that the second conjunct is a full-fledged clause, syntactically and semantically. This claim is supported by numerous lines of reasoning and various constructions, too numerous to outline here. One of the simplest and most elegant pieces of evidence is found in the constructions in 7 (examples 7a,b appear on page 186 of my book).

Unless we assume that there is an object gap in 7a at the left edge, there is no way to interpret the construction grammatically. In 7b, the ungrammaticality stems from the fact that both an object gap and an adverb may not occur in the initial position of the second conjunct. Given this evidence for a left-edge object gap, it is theoretically consistent in a Minimalist approach to assume that there is a subject gap in 7c. The assumption that there is no gap results in inconsistent, inelegant syntactic theory. This argument is in part theory-internal, but it also follows Occam's Razor, a theory-independent principle. Assuming that only 7c has no left-edge gap is not only (theory-independently) inelegant, it also creates theory-internal problems from a generative standpoint, especially in regard to the syntax of West Germanic, a point that I argue extensively in chapter 5. Osborne does not comment on this extensive argumentation and therefore misrepresents my work. I suspect his reason for overlooking my argumentation and data is motivated by his own work in Dependency Grammar, where it is more advantageous to assume that constructions like 7c have no subject gap.

Osborne further discusses my treatment of the very problematic construction in 8.

He states: “in order to overcome the problem of [8], the derivation […] must first select a set of paintings as the object of each conjunct, delete these objects, and then replace them with the same set of paintings.” The derivation from page 259 of my book is given in 9.

It remains a mystery to me how Osborne could make the statement above based on this derivation and the extensive discussion of Right Node Raising (RNR) in chapter 4, on which this derivation is based. The only difference between this derivation and the derivation of any RNR construction is the merger of the element the same in 9; in all of them one set of redundant elements is not realized in Phonological Form (PF) (indicated by the strikethrough), that is, there is no deletion and replacement operation as Osborne claims.

Further on, Osborne claims that my proposal is not able to account for the constructions in 10.

He argues that because the gaps in these constructions do not occur in C° or T° as required in my theory of Gapping, my proposal cannot account for them. Again, it appears I should have been more redundant in my discussion of Gapping where I outline the properties of Gapping constructions, which are incidentally the same properties assumed widely in the generative literature. Specifically, Gapping involves only a finite verb and optionally an adjacent complement. The constructions in 10 do not meet these requirements, and thus my proposal for Gapping does not apply to them. My proposal for coordinate structures outlined in chapter 3 does, however, account for such constructions. In 10a, the second conjunct can be merged (after the derivation of the first conjunct is complete, that is by phase) within the scope of the infinitive zu sagen for the required interpretation. In 10b, the derivation is almost identical to the derivation of constructions covered in section 4.2 (chapter 4). Although it is excusable that this was not recognized, it should have been clear to Osborne from my definition that 10b is not a case of Gapping.

The same type of problem occurs in his discussion of constructions that clearly fall under Verb Phrase Deletion (VPE), to which I devote an entire section. My definition of VPE follows standard assumptions. Osborne faults me for not including them in my discussion of Gapping. Obviously his criticism is not really directed here at my proposal but rather at assumptions about how forms of coordinate ellipsis should be classified. Is this appropriate in a review of my work? Again, it seems to be an indication of his personal bias, a negative approach to the review, and not of a careful critique of my work in the context of other generative research.

Osborne claims that my proposal for Gapping cannot account for 11 because, as he states, “there is an unlicensed gap preceding the initial remnant.”

In chapter 4 (section 4.5.3), I present constructions in examples 180–182 that are configurationally identical to 11 and account for them using my proposal for Left Edge Ellipsis, which explains why these constructions do not have the characteristic prosodic features of Gapping constructions.

In his concluding comments, Osborne presents 12 as a construction with two possible interpretations, one of which, he argues, my proposal cannot capture.

Example 12 can be interpreted—depending on the intonation used—as either a yes/no question, or as an information question with the answer “coffee”/“tea.” I do not explicitly discuss such constructions in my book, but an account of them follows logically from the principles of my proposal for coordinate ellipsis in which the interface of syntax and prosody captures the two interpretations of 12.

It is understandable and excusable that this point was overlooked by Osborne, for it requires a synthesis of the various aspects of my proposal for coordinate ellipsis that usually comes only after working with a proposal of this sort over an extended period of time. Not excusable, however, are other major criticisms in his review that reflect a superficial reading and some level of misreading due to theoretical bias. If a reviewer wants to level extensive criticism, it behooves him/her to read carefully and extensively (and thereby overcome biases).