Norvin Richards's Uttering trees presents two novel, albeit very different ideas on linear order and syntactic structure, each of which is developed in its own separate, large chapter. On the surface, these two ideas appear rather incongruously placed together into a single monograph. To draw a musical analogy, however, Mahler's Eighth Symphony consists of two seemingly disparate movements, that, upon more thoughtful inspection, weave a common theme throughout. Likewise, both of Richards's ideas address important and unresolved issues concerning the syntax–phonology interface. The first idea is that a given phase cannot contain two categories of the same type or else they cannot be linearized. From this he explains a host of recalcitrant problems in syntax. The second idea concerns wh-movement, and why some languages have it and some do not. Richards proposes a closeness constraint between a wh-phrase and its associated C(omplementizer) head such that the two must appear in the same ‘minor phrase’ (148–153). These two proposals together address issues of how a hierarchical phrase structure is transformed into a linear sequence.

The book opens with a brief ‘Introduction’ in Chapter 1, which summarizes the core proposals in two pages. Chapter 2 presents Richards's theory of ‘Distinctness’. He proposes that within a given phase, there can be no two nodes with the same label. Assuming that individual phases are sent to Spell-Out individually, only the set of labels within a phase is evaluated for linearization. Adopting Kayne's Linear Correspondence Axiom, Richards proposes that the derivation crashes if a linearization statement of the form <α, α>is generated. This simple and elegant condition is the core of Distinctness.

He illustrates with the following puzzling asymmetry in English multiple sluicing constructions:

1. (1)

   1. (a) I know that everyone danced with someone, but I don't know who with whom.

   2. (b) *I know that everyone insulted someone, but I don't know who whom.(3, exx. (1)–(2))

Multiple sluicing constructions in English cannot contain two Determiner Phrases (DPs). Richards attributes this to Distinctness. The relevant phase in (1b) contains <DP, DP>, which cannot be linearized. The acceptable version contains <DP, PP>, with distinct labels, hence linearization can proceed.

Richards adduces several other lines of evidence from a wide range of languages to support his thesis, including multiple exceptives, multiple DP-internal arguments, causatives, DP-predicates in verb-initial languages, quotative and locative inversion. It should be noted that many of these phenomena have been difficult to account for in previous frameworks but receive a unified treatment under Richards's Distinctness proposal. Richards adds further data to the mix that were previously explained as linear adjacency violations, as in the following paradigm, first discussed by Ross (Reference Ross1972) .

1. (1)

   1. (a) It continued to rain.

   2. (b) It continued raining.

   3. (c) It's continuing to rain.

   4. (d) *It's continuing raining.(23, ex. (37))

Richards points out, however, that an intervening wh-trace saves the construction.

1. (2) the children [that I was watching t playing](p. 23, ex. (38))

Richards goes on to discuss even more data, which show that strict linear adjacency is not at play here. A well-known constraint on quotative inversion is that it is not available when an internal object without a preposition is present. Richards argues that this is not a simple adjacency effect. Rather, (3c) is ungrammatical due to the presence of two DPs in the same phase (although it should be noted that adverbs do not normally precede objects in English, casting doubt on the relevance of the example).

1. (3)

   1. (a) ‘It's cold,’ said John to Mary.

   2. (b) *‘It's cold,’ told John Mary.

   3. (c) *‘It's cold,’ told John sadly Mary.(p. 13, ex. (22))

These are not the only phenomena discussed in this data-rich section; however, space prevents a fuller discussion of the empirical facts here.

The simplicity of this proposal, that no two nodes within a given phase may have the same label, faces some challenges that require additional refinement. While English does not allow multiple sluicing with two DPs, Japanese does. Thus, forms such as but I forget who what are available in Japanese while unavailable in English. Richards argues that the rich overt case marking on nominals in Japanese is what allows such forms to be acceptable. Thus, for the purposes of linearization, Japanese can distinguish between two distinct Case Phrases, KPnom and KPacc. Interestingly, such forms are not available in Japanese in double nominative constructions, which Richards attributes to the presence of two KPnoms in the same phase. Readers familiar with Richards's work will recall his earlier work on multiple wh-movement. Again, it appears as though multiple wh-movement languages allow two DPs in the same phase. However, Richards presents data from Bošković (Reference Bošković2001) that turn out to support Distinctness. Identical DPs are not permitted in such constructions. The two raised wh-phrases must be distinct in some way.

1. (4)

   1. (a) Kokogavidi?Serbianwhowhomsees‘Who sees whom?’

   2. (b) *Štaštauslovljava?whatwhatconditions‘What conditions what?’(50–51, exx. (104)–(105))

Next, Richards discusses how languages might avoid Distinctness violations. Some of these are discussed in previous sections, such as the addition of a preposition (John loves apples vs. John's love ofapples). It is in this section that Richards refines Distinctness even further in two ways. First, he parameterizes what features on nodes count as distinct. Second, he makes Distinctness a dynamic constraint.

Parameterized sensitivity to features for tallying Distinctness violations was alluded to above. In Japanese, the value of Case on a KP is pertinent to the evaluation of Distinctness, while in English, two KPs in the same phase will trigger a Distinction violation regardless of the value of Case. Richards shows cases where a difference in ϕ-feature sets is enough to alleviate a Distinctness violation. Thus, two feminine DPs in the same phase cause the derivation to crash, but a feminine DP and a masculine DP in the same phase do not.

Richards further shows that Distinctness is a dynamic constraint. Distinctness is clearly not an inviolable constraint on structure since violations of it can be alleviated by movement. Taking inspiration from Moro's (Reference Moro2000) Dynamic Antisymmetry, he develops a derivational principle of Distinctness, which states that violations of it must be as brief as possible.

This chapter makes far-reaching claims on the constraints of structure and how trees are treated at the PF (Phonological Form) interface. Mustering well-researched facts from an impressive array of languages around the world, including some for which very little prior work exists (such as the Ethiopian Semitic language Chaha), Richards makes a bold claim that generates clear empirical predictions to stimulate future work.

Chapter 3, ‘Beyond strength and weakness’, addresses a perennially difficult question on wh-movement. Why do some languages have it and others do not? Again, Richards proposes a novel solution. He argues that a wh-phrase and its associated complementizer must be separated by as few prosodic boundaries as possible. There are essentially two ways to ensure that C and wh appear in the same prosodic phrase: either wh-movement or prosodic levelling. Starting with a discussion of various studies on the prosodic properties of wh-questions in Japanese, Richards notes a levelling effect between the wh-phrase and the final C in Japanese questions. Prosodic levelling results from creating a larger ‘minor [prosodic] phrase’ from one edge of the minor phrase containing the wh-phrase and any other pre-existing edge of a minor phrase (154ff.). If such a domain can be made to include C, then no movement is necessary. This larger minor phrase is diagnosed by prosodic levelling as discussed.

To account for the cross-linguistic variation in wh-constructions, Richards proposes the following two parameters. The first is derived from the typological literature. The second is derived from studies on the syntax–phonology interface, which Richards draws on heavily for this discussion.

1. (5)

   1. (a) Complementizers either precede or follow their complements.

   2. (b) Phrase boundaries are either to the left or right of certain XPs.

This gives rise to a four-way typology of languages, which Richards goes on to exemplify. Thus, if a language has final C and minor phrase boundaries at the left edge of XP, the right edge of the larger minor phrase can be selected so as to include C. Richards argues that this is the case for Japanese; thus, no movement is required. Languages that are C-initial and place minor phrase boundaries at the left edge of XP will have obligatory wh-movement. Space precludes a fuller discussion of the possibilities; however, Richards gives a clear and thorough discussion of all four possible types, with ample empirical evidence.

This proposal makes several strong predictions, one of which appears to be problematic. It predicts that C-final languages should never have wh-movement, as this only moves the wh-phrase further from C. Movement that serves to increase the distance between the wh-phrase and C appears to be incompatible with Richards's claim that the sole motivation for wh-movement is to place these two elements in the same minor phrase. The typological literature clearly suggests a correlation between C-final languages and wh-in-situ; however, Bruening (Reference Bruening2007) unequivocally shows that such a correlation is not valid. Crucially, he shows several examples of C-final languages with overt, leftward wh-movement. One possible way out of this is to argue that the final C-particles are actually speech-act markers that agree with illocutionary force, and that the real C head is at the left edge.

Richards admits further unexplored areas of research, such as multiple wh-movement, which does not seem quite amenable to the approach that he advocates. It is possible, though, that multiple wh-movement is a distinct phenomenon. Another possible avenue of research is to extend Richards's proposal to other kinds of A-bar movement, and perhaps even other kinds of A-movement. Why do some languages leave the subject in the specifier of vP while others raise it? Can the presence of transitive expletive constructions in Icelandic be explained by its prosody? Richards's proposal sets the stage for a research program of considerable scope.

The book ends in Chapter 4 with a two-page ‘Conclusion’. In terms of structure and design, this monograph has the distinct appearance of a collection of two very long, independent contributions, each of which would be too long as a journal article. Taken together, though, they clearly exhibit a level of theoretical depth and empirical breadth worthy of a book-length manuscript. The two topics are not as incongruous as they may superficially appear (as I alluded to above). Indeed, they both offer novel, well-thought-out solutions to some nagging problems of syntax, relating them to issues of linearization at the PF-interface. Both chapters lead the reader carefully through a basic description of the proposals and proceed to apply these proposals to various sets of data to refine the approaches and point out areas for future research.

It is always easy to poke holes in a new research program that is founded on a simple, elegant principle. Such proposals, however, make for better science, as the holes in them are more easily spotted than in weaker, more eclectic proposals. The reader should thus take the following criticisms as suggestions for future research and refinement of the ideas presented in this monograph.

Regarding Distinctness, there seem to be some constructions which contain equivalently labelled phrases in the same domain. German container-constructions come to mind.

1. (6) eine Dose Cola

   one can cola

   ‘a can of cola’

Within the DP, there appear to be two noun phrases. Other problematic areas are in the domain of compound formation (blue-green, a mother–daughter relationship, investment banker). Richards argues that two lexical roots in the same phase are not problematic for Distinctness since they can be distinguished by their content rather than by their label, so one might be tempted to explain these forms away as root–root compounds. Given the clear presence of categorizing morphology on these forms, however, it is clear that they are not merely concatenated roots. Serial verb constructions are also problematic, as they presumably contain sequences of verb phrases within the same phase. Readers will undoubtedly find further apparent counter-examples to Distinctness in the languages on which they work.