Book announced at https://linguistlist.org/issues/35-765

Title: Coordinate Structures
Series Title: Elements in Generative Syntax
Publication Year: 2023

Publisher: Cambridge University Press
http://www.cambridge.org/linguistics
Book URL: https://cambridge.org/9781009326636

Author(s): Ning Zhang

Reviewer: Dennis Ott

SUMMARY

This slim monograph is the second entry in Cambridge University Press’s “Elements” series. At a mere 67 substantive pages, and like the two other entries in the series published at the time of writing, it is essentially an extended research article, structured into “sections” rather than chapters.

Section 1 opens with general introductory remarks on coordination and argues that—pace Johannessen 1998 and other works—coordinate structures do not have a distinct, inherent category (&P, ConjP, or the like); rather, their category is derivative, determined by that of the structurally prominent conjunct. Zhang’s central thesis is that coordination arises as a result of stepwise complementation and categorization: in a first step, the coordinator (& for now) is merged with an “internal” conjunct Y to form [ & Y ]; in a second step, this head–complement structure is adjoined to an “external” conjunct X to form [ X [ & Y ]], of category X. I will refer to this core claim of the monograph as the modification analysis of coordination (MAC).

Section 2 seeks to buttress the MAC by highlighting apparent signs of structural asymmetry between conjuncts, indicating that the external conjunct c-commands the internal conjunct. Evidence cited involves Q-binding and NPI-licensing. Furthermore, Zhang argues that only the external conjunct is truly integrated into the higher clausal structure, as evidenced by the fact that it, not the internal conjunct, must satisfy externally imposed selectional restrictions (e.g. of adpositions embedding complexes coordinating nominal and clausal categories).

The MAC is then spelled out in some further detail. Coordinators are argued to be functional items that assign no theta-roles but select complements. Zhang generalizes over different types of coordinators as belonging to an abstract category J; she consequently refers to the head–complement configuration [ J Y ] formed by merging J with the internal conjunct Y as a “J-set”. This J-set in turn merges with the external conjunct X to form [ X [ J Y ]], of category X(P).

J is claimed not to be a bona fide syntactic category, but a root-like element; as a result, the J-set is neutral between argumental and predicative uses and remains unlabeled. This, Zhang argues, is what motivates merging the J-set with an external conjunct: absent a label for the J-set, the external conjunct provides the category of the overall coordinate complex. The J-set itself is an adjunct to the external conjunct and as such syntactically optional, an assumption shared with Munn 1993.

Zhang adduces further support for the MAC by pointing out a number of properties shared by J-sets and adjuncts, such as their optionality, iterability, and positional variability (the J-set can precede or follow the external conjunct in linear order, with additional phonological factors deciding the positioning of the coordinator).

As for the interpretation of coordinate structures, Zhang’s claim is that the interpretation of a J-set [ J X ] depends on the interpretation of X as a predicate, entity, or proposition; in the general case, the external conjunct will be of the same semantic type, although cases of so-called asymmetric coordination are argued to fall under this analysis as well.

Section 3 goes deeper into the morphosyntax of coordinators and develops the MAC’s core idea that J is a general modification marker or linker, rather than a dedicated coordinator. J is explicitly likened to Tagalog-style linkers and Mandarin’s modification marker de and assimilated to these “dependency markers” on the grounds of several shared properties: all are functional elements (non-theta-assigners); all lack an inherent category; all link two elements; all can be null; all can be proclitics (e.g. English) or enclitics (Japanese) and are variable in their positioning; in addition, coordinators and linkers share a common form in some languages. The net outcome are directly analogous representations for modification and coordination:

[ VP [ J VP ]] (e.g. “read the paper and watch the movie”)

[ VP [ J AdvP ]] (e.g. “read the paper quickly”, with J null)

Additional coordinators (as in French “et XP et YP”) are said to be correlative “focus markers” (echoing De Vries 2005); the true exponent of J, Zhang claims, is always the one occurring with the internal conjunct.

The idea that the linker (c-)selects the internal conjunct is supported by the observation that many languages have different coordinators for the coordination of different categories (e.g. nominal vs. clausal).

The section also discusses coordinate sequences of an unbounded number of conjuncts, which Chomsky (1961,2013) calls “unstructured coordination”, as in “noodles, (and) eggs, (and) lettuce, and juice”. Against Neeleman et al. (2023), Zhang argues that such sequences do in fact realize nested J-set structures enabled by nonexhaustive coordinators and are thus not unstructured at all. In English/Mandarin-type languages, such coordinate complexes feature either a coordinator in each conjunct or else a single overt coordinator next to the final conjunct.

Zhang further argues that with three or more conjuncts, “subgrouping” (recursive coordination) is impossible unless each conjunct occurs with an overt coordinator. (This generalization, Zhang’s (71), does not seem to be empirically correct; perhaps it can be read charitably as an idealization covering the examples to be discussed.) That is, the expression “Tom and Dick and Harry” can be assigned either of two structures, each giving rise to multiple readings:

[ [ Tom [ J Dick ]] [ J Harry ] ]

[ Tom [ J Dick [ J Harry ]] ]

By contrast, no such subgrouping is possible for “Tom, Dick and Harry”, where “Dick and Harry” is said to be a “complex J-set” [ Dick [ J Harry ]], to be categorized by the NP Tom, rather than a coordinate structure; as a result, “these two conjuncts do not form a group semantically” (32); see Borsley 2005 for discussion of these contrasts. The idea seems to be that J can in principle have any number of specifiers, and that this structuring around a unique J contrasts with recursive coordination (nested J-sets) enabled by iteration of J.

Section 4 seeks to adduce further evidence for the MAC’s central thesis that coordination is a subtype of modification, and that the composition of coordinate structures does not require any construction-specific operations.

Zhang rejects the idea that all coordination is at the clausal level (as prevalent in early theorizing and more recently revived in Schein’s 2017 work). J is neutral with regard to the categories of its dependents: instead of a rigid constraint requiring “coordination of likes”, Zhang, building on Kehler 2002 and her own earlier work (Zhang 2010), argues for a constraint imposing “coherence” on the coordinate complex, which is defined in terms of the notions “relatedness” and “resemblance”. This constraint is said to “reflect the economy in processing, rather than a constraint on the operations in building a coordinate complex” (41) and assumed to supersede the Coordinate-structure Constraint (CSC); Zhang provides various arguments for the syntactic mobility of conjuncts and their elements.

Section 5 considers derivative constructions: ATB-movement and related phenomena. Zhang distinguishes “I-ATB” with identical gaps, “R-ATB” with respectively-type interpretations, and S-ATB with additive interpretations. I-ATB is argued to involve an anaphoric pro in non-initial conjuncts (“Who did John see t and Mary kiss pro?”), i.e. extraction proceeds asymmetrically from the initial conjunct. R-ATB is claimed to involve repetitions and ellipsis (“How many students did John see and <how many students did> Mary kiss, respectively?”), rendering such configurations parallel to sequences of questions with obligatorily non-identical gaps. S-ATB is claimed to be the result of movement of a null operator in each conjunct, which derives a predicate; the multiple derived predicates then combine into a compound predicate of which the left-edge XP is a kind of subject (“How many books [ OP did John borrow t ] and [ OP Mary steal t in total ]?”), yielding the additive reading. Alternative analyses—forking chains, sideward movement, multidominance—are briefly considered but rejected more or less out of hand.

Zhang furthermore addresses what she calls “heterofunctional coordination”, as in “How and why did she kill him?” (or the Russian equivalent of “Who and whom saw?” = “Who saw whom?”), argued to involve sideward movement and an expletive coordinator, and “split-argument constructions” with symmetrical predicates (“John met Mary” = “John and Mary met”) as well as modifier sharing/split antecedents, argued to derive via conjunct movement from an underlying “hydra” (“a man and a woman who knew each other”). See Van Craenenbroeck & Johnson forthcoming for an almost identical analysis of symmetrical predicates.

Lastly, Zhang proposes a sideward-movement analysis of what Bošković (2022) terms “non-ATB ATB” (as in “How many books and how many magazines did John write and Mary read?”), which again involve a derived coordination in operator position, Zhang suggests. The analysis proposed seems to be virtually identical to Bošković’s.

Section 6 concludes the essay by reiterating the central tenets of the MAC:

[1] There is no dedicated functional head &, only a general linker J variously realized as a coordinator or a modification marker, taking a complement and a specifier.

[2] No specialized operations are involved in the construction of coordinate complexes.

[3] Coordination is an instance of modification. The differences between coordination and other modification configurations are entirely due to the interpretive relation (symmetric/asymmetric) between the XPs related by the linker J and lexical properties of J.

EVALUATION

Ning Zhang’s “Coordinate structures” is a highly relevant contribution to the theoretical debate surrounding coordination and an ambitious attempt at assimilating the phenomenon to the better-understood mechanisms of modification. The general notion that coordinate complexes are not endocentric but categorized by the prominent conjunct seems to be fundamentally correct (see also Borsley 2005, Chomsky 2013). Given its simplicity, the MAC could plausibly be considered a null hypothesis.

As usual, questions remain and new problems arise.

The book repeatedly makes sweeping generalizations based on rather casual examination of a small number of (usually English) data points; for instance, the evidence presented for structural prominence of the external conjunct over the internal conjunct based on binding is more controversial than Zhang acknowledges (see Ke et al. 2025 for recent critical discussion). The proposal would have strongly benefitted from a more in-depth discussion of these fundamental properties of coordination.

The nature of J remains somewhat obscure. On the one hand, as Zhang makes clear, the analysis depends crucially on its NOT being a syntactic category (since this would label coordinate complexes as JPs); yet, at the same time, the claim that J groups together a range of LIs is equally central to the proposal. It is hard to see, for this reviewer at least, what this means if not that J is, in fact, a syntactic category. Furthermore, J is claimed to be both a functional category and root-like in its inability to effect labeling. Note that this view seems to entail labeling of the J-set by the internal conjunct in head–head coordinations (which seem to be real: Borsley 2005), a possibility Zhang does not discuss. These apparent contradictions are not resolved in the monograph.

The analysis of “unstructured” coordination as structured similarly raises questions. Modifiers, and categories quite generally, can be added to expressions in two ways, hierarchically (“the oldest blue house”) and sequentially (“the oldest, blue house”). It remains unclear if Zhang’s analysis of “unstructured” coordination as involving either a “complex J-set” constructed around a single J or recursive J-set formation can elucidate these analogous differential modes of composition. A related worry is that the formation of complex J-sets to avoid recursive coordination in cases such as “Tom, Dick and Harry” requires the stipulation that an XP merged to a J-set (“Dick”, in this case) can fail to categorize it. Why this is remains unclear.

The discussion of ATB constructions feels rather like an add-on to the discussion than an integral part of it; in this reviewer’s estimation, the monograph would have benefitted from omission of this part in favor of a more detailed exploration of the MAC and its ramifications. Zhang’s assertion that “all three types of ATB constructions are derived without any coordination-specific syntactic operations” (49), repeatedly emphasized throughout, is borne out by the discussion only in an extremely narrow sense, since the analyses presented rest on various construction-specific stipulations (e.g., the exceptional availability of radical pro-drop in I-ATB configurations in grammars that do not generally permit it).

Lastly, it remains unclear what, if anything, remains of the CSC. Zhang asserts that “conjuncts may move, and elements may also move from conjuncts” (65), and the explanatory burden of canonical CSC effects appears to be shifted entirely to the coherence constraint. Whether or not this is the right move to make can only be evaluated once the latter is elaborated in greater detail.

Despite these drawbacks and remaining questions, Ning Zhang’s essay is a valuable contribution to a long-standing debate in syntactic theory and commendable for its scope, clarity, and dedication to simplification of the theory. The author’s hope, expressed in the monograph’s closing sentence, that it will “serve as a springboard to further research on coordinate structures and syntactic theories”, is likely to be fulfilled.

REFERENCES

Bošković, Ž. 2022. On the limits of across-the-board movement: distributed extraction coordinations. Philosophies 7. https://doi.org/10.3390/philosophies7010010

Chomsky, N. 1961. On the notion ‘rule of grammar’. In R. Jakobson (ed.), Structure of Language and its Mathematical Aspects, 6–24. Providence. RI: American Mathematical Society.

Chomsky, N. 2013. Problems of projection. Lingua 130.

van Craenenbroeck, J. and K. Johnson. Forthcoming. Sloppy symmetry under ellipsis. The Linguistic Review.

Johannessen, J.B. 1998. Coordination. Oxford University Press.

Ke, A.H., A. McInnerney, and Y. Sugimoto. 2025. Lack of c-command in coordinate structures: evidence from binding. Natural Language & Linguistic Theory, online first. https://doi.org/10.1007/s11049-025-09662-8

Kehler, A. 2002. Coherence, reference, and the theory of grammar. Stanford, CA: CSLI Publications.

Munn, A. 1993. Topics in the syntax and semantics of coordinate structures. PhD dissertation, University of Maryland.

Neeleman, A., J. Philip, M. Tanaka, and H. van de Koot. 2023. Subordination and binary branching. Syntax 26.

de Vries, M. 2005. Coordination and syntactic hierarchy. Studia Linguistica 59.

Schein, B. 2017. And. MIT Press.

Zhang, N. 2010. Coordination in syntax. Cambridge University Press.

ABOUT THE REVIEWER

Dennis Ott is associate professor of linguistics at the University of Ottawa, Canada. His research focuses on various areas of syntax, including unbounded dependencies, connectivity, and ellipsis.