This book offers a comprehensive account of German syntax within a novel theoretical approach. Departing from a fairly traditional, feature-driven, and phase-based version of Chomsky’s (Reference Chomsky1995) Minimalist Program, Müller introduces Remove as a syntactic operation alongside Merge and Agree. Remove works as a mirror-image of Merge; while the latter builds syntactic structure, the former removes it. Müller convincingly shows that this operation can explain a number of syntactic phenomena which are difficult to handle within syntactic theory because they involve conflicting structure assignment, such as (long-distance) passives, restructuring configurations, and complex prefields in German. In addition to pursuing this analysis, the book also provides a comprehensive and very pedagogical introduction to German minimalist syntax.

The book consists of eight chapters, an extensive bibliography, and a useful index. Following the order of these chapters, I will first provide a summary of Müller’s minimalist syntactic machinery (Chapter 1), before we have a closer look at the novel operation Remove (Chapter 2) and its application throughout the book (Chapters 3–7).

In the first chapter, Müller lays out the basics of structure building in German within his version of the minimalist framework. An architecture of the grammar is presented that can be placed somewhere in between purely lexicalist approaches, which assume the lexicon to be a separate, generative component of the grammar (Chomsky Reference Chomsky1995), and non-lexicalist approaches such as Distributed Morphology, according to which the traditional lexicon is distributed across the syntactic component (see e.g. Harley & Noyer Reference Harley and Noyer1999). On Müller’s approach, lexical items are associated with a phonological representation, an argument structure, and morphosyntactic features as they enter the syntactic derivation, in line with lexicalist assumptions, but he assumes that ‘the lexicon is not a generative component’ (p. 7) and that there exists a pre-syntactic morphological component (i.e. the numeration) where lexical items are associated with non-inherent morphosyntactic features.

In a standard minimalist vein, such features are the driving force behind all syntactic derivations. The basic operations in Müller’s syntax are Merge and Agree, in addition to his invention Remove. Merge is triggered by structure-building features [●F●] that are ordered on lexical items and syntactic heads in the numeration, as in (1), and deleted once Merge between two elements with matching features has taken place; see the derivation of a basic, transitive vP in (2). Alternatively, Merge can be triggered if this results in semantic modification, i.e. adjunction (see the Economy constraint on Merge on p. 31).

Agree is triggered by probe features [*F*] looking for matching goal features. Interestingly, Müller assumes that the probe can look either upwards or downwards in the structure for a matching goal, meaning that this operation is not limited to the traditional notion of downward Agree (following Chomsky Reference Chomsky, Martin, Michaels and Uriagereka2000, Reference Chomsky and Kenstowicz2001) or the somewhat newer variant of upward Agree (see e.g. Zeijlstra Reference Zeijlstra2012). Importantly, both Merge and Agree (and, as becomes clear in Chapter 2, Remove) are subject to locality restrictions. Specifically, they are constrained by the Strict Cycle Condition (see p. 2) and the Phase Impenetrability Condition (see p. 62), whereby vP and CP are assumed to constitute phases.

Furthermore, Müller argues that a finite German clause contains both a TP and a CP layer. Movement into these higher domains of the clause via internal Merge is brought about by abstract (and generally optional) structure-building features that encode different movement types. Importantly, features causing movement are ordered below features triggering External Merge; logically, a piece of structure needs to be introduced in the syntactic structure before it can be moved.

With these basics in place, Chapter 2 introduces the syntactic operation which is at the heart of the rest of this monograph: Remove. This operation is also feature driven, by a third type of features [–F–], which are ranked below Merge (structure-building) features. Müller argues that such an operation is ‘to be expected for reasons of symmetry alone’ (p. 97), acting as the mirror image of Merge. Just like Merge, Remove obeys locality restrictions and can apply recursively. Furthermore, it can apply to either heads or phrases, as Müller demonstrates pedagogically for many different syntactic constellations on pp. 99–106. (3) is an example of Remove applying to a phrase in complement position, and (4) shows how a head is removed. Importantly, removal of a head only takes away the head Y and its highest projection YP, i.e. the YP shell. If the phrase YP contains a complement WP or a specifier ZP, this is reassociated with the dominating projection; see (4b).

To avoid problems for semantic interpretation, a general recoverability requirement is assumed: Removal is only possible if the denotation of the deleted item is recoverable from the syntactic context or a default interpretation is inserted in the semantic component (p. 110). Parallel to how Merge takes an item from the workspace to build structures, Remove puts an item back into the workspace, from which it can be reinserted into the structure as an adjunct at a later point (this is what happens when the demoted agent of a passive occurs as a by-phrase, as argued in Chapter 3; see below).

From a conceptual point of view, one may be sceptical of the idea that syntactic structure is first built and then removed again – only to, in some cases, be put back into the syntax at a later point. As Müller points out, the idea that structure can be built and removed again is not new, but has its roots in early generative grammar (see Ross (Reference Ross1967) on Tree Pruning or Chomsky’s (Reference Chomsky1981) S-Bar Deletion in ECM constructions). It could be questioned whether such an approach is in line with current minimalist principles of economy or how it can ‘make sense’ on a conceptual level, as Müller himself discusses on p. 108 f. However, the reasoning behind Remove is quite convincing, and this analysis has an impressive empirical coverage, as becomes clear in Chapters 3–7.

The empirical basis for Remove is the contradictory behaviour of certain constructions where a part of the structure appears to be both present and absent at the same time, such as the external argument of passives or the TP and CP domains of the embedded clause in a restructuring configuration, as we will see in more detail below. As Müller shows, all these phenomena have received different treatments before, but these vary in type, they often involve analyses without independent motivation, and – crucially – they fail to capture an important new generalisation that Müller draws across these empirical phenomena: a syntactic element with this contradictory behaviour is consistently accessible from positions below it but inaccessible from positions above it. This can be explained by an analysis whereby a specific part of syntactic structure is first built and available for operations involving elements that it c-commands (or, for the external argument in passives, m-commands), before it is removed and consequently becomes unavailable for operations involving elements that are introduced after its deletion. To use Müller’s frequent expression, the constituents affected by Remove have a short life cycle.

The remainder of the book provides Remove-based analyses for a range of phenomena in German syntax. In each case, Müller first shows how the phenomenon is difficult to handle within standard approaches, before he goes on to show how a Remove-based analysis can be implemented. We will have a look at Müller’s argumentation in some detail for passives in Chapter 3, while Chapters 4–7 will only be briefly summarised below.

First, this analysis is applied to passives in Chapter 3 to account for the conflicting behaviour of the demoted external argument. Applying a battery of tests, Müller shows that the external argument is accessible for operations involving elements in its m-command domain, but inaccessible from higher positions. To mention just a few examples provided in the chapter (pp. 115–135), it is accessible as a controller of PRO (via Agree) in an adjunct clause, secondary predicate, or complement clause, and it can bind reflexives and reciprocals. At the same time, it cannot be bound by a quantified item in a higher clause, and it does not block anaphoric binding from above, which external arguments usually do. Müller proposes an analysis whereby the external argument of a passive clause is first introduced in Spec,vP, but because passive v has the feature [–D–], it is removed again before the TP and CP domains of the clause are merged. Müller argues that such a [–D–] feature is the core property of passivisation not just in German, but cross-linguistically.

The external argument thus has a short life cycle, during which some operations can take place, such as control and binding of lower elements, but no operations involving higher elements. This means that the timing of the various operations, i.e. the ranking of different features on syntactic heads, is key here. However, after having ‘died’ via Remove, the external argument can be ‘resurrected’ again (to use Müller’s metaphors). The assumption is that Remove puts the external argument back into the workspace, from which it can be retrieved in the only way that is possible without structure-building features, i.e. as an adjunct von-/by-phrase, which will be accessible for subsequent syntactic operations.

Arguably, the extension of this analysis to adjectival passives, or the Zustandspassiv, on p. 156 ff., is a bit questionable. As is well documented in the literature, there is no evidence for an external argument in the Zustandspassiv (e.g. Die Geißlein sind versteckt / lit. The goatlings are hidden), even though we would expect one based on the verb’s (verstecken / hide) argument structure. Müller argues that the (potential) external argument in a Zustandspassiv is removed from the workspace (or, more specifically, moved from ‘the active part of the workspace into the inactive part’ (p. 158)) before it has a chance to enter the syntax; this is an instance of external rather than internal Remove, again in parallel to Merge. Thus, it is not accessible to operations like control and cannot re-emerge as a von-/by-phrase. Continuing the life and death metaphors, Müller (somewhat apprehensively) calls this an ‘abortion’ in fn. 34 of Chapter 3. However, keeping such an argument in the workspace when it is not needed for syntactic or semantic purposes seems unnecessary. Furthermore, as Müller notes on p. 159, this analysis cannot explain why the Zustandspassiv is incompatible with temporal adverbials and at the same time available for un-prefixation, something which analyses that consider such constructions to be adjectival instead – either due to a transformation in the lexicon (see Wasow Reference Wasow, Culicover, Wasow and Akmajian1977) or because a verb is stativised in the syntax (see Alexiadou et al. Reference Alexiadou, Anagnostopoulou and Schäfer2015) – can account for.

Chapter 4 deals with restructuring in control infinitives, as in (5).

Clauses containing a restructuring infinitive have both monoclausal and biclausal properties, as a vast literature on the topic has documented. Again, Müller convincingly connects these properties to downward accessibility and upward inaccessibility. He proposes a uniform analysis whereby all control infinitives start out as CPs, but in the case of restructuring, Remove applies cyclically to the C and T heads. This accounts for their clausal properties at a certain point of the derivation as well as their non-clausal properties later in the derivation.

In Chapter 5, Müller neatly combines the two analyses from Chapters 3 and 4 to account for long-distance passivisation, as in (6):

An analysis is proposed whereby the embedded infinitive is first passivised (via removal of the external argument) before the CP and TP layers are removed (= restructuring). In a third operation, the vP layer of the infinitive is removed. The latter operation explains the nominative marking of the original object (der Traktor) in the infinitive: Without vP, the object cannot receive accusative case and must instead get nominative from matrix T. Appealingly, the output is a very small representation that is ‘almost indistinguishable from Wurmbrand’s (Reference Wurmbrand2001) proposal according to which [these structures] involve embedding of a bare VP’ (p. 236).

Chapter 6 is concerned with complex prefields, where more than one constituent appears to precede the finite verb of a V2 clause, as in (7):

Syntactic tests indicate both that the phrases in such a prefield constitute a single constituent, i.e. the VP, and that this is an instance of multiple fronting of several constituents, i.e. a true V2 violation. Again, Müller’s analysis can account for this conflicting evidence: First, a VP with an empty V head is fronted to Spec,CP, yielding single constituency effects. Then the VP shell is removed, which involves removing the V head and VP layer, whilst reassociating its specifiers (i.e. the constituents of a complex prefield) with specifier positions of the CP. This gives rise to multiple constituency effects.

Finally, in Chapter 7, Müller makes the case that Remove can explain a range of other syntactic phenomena as well: bridge verbs, applicatives, null objects, pseudo-noun incorporation, nominal concord, and ellipsis. The analyses in this chapter are not as detailed as in the previous chapters and can be considered sketches for future research. This chapter shows that taken together, the empirical coverage of the Remove analysis is impressive.

In his conclusion in Chapter 8, Müller points to an important potential for his Remove-based framework: It may be able to reconcile two opposing strands of syntactic research, i.e. cartographic approaches arguing for increasingly complex syntactic structures (see e.g. Rizzi & Cinque Reference Rizzi and Cinque2016) and approaches arguing for a simpler syntax, such as Culicover & Jackendoff (Reference Culicover and Jackendoff2005). With Remove, both positions can be correct: ‘There can initially be structures of substantial abstractness and complexity, which can then subsequently be reduced to simple representations that are quite surface-oriented’ (Müller, p. 331).

However, it remains to be seen whether proponents for a simpler syntax will in fact find the Remove-based approach appealing, as it is certainly not without complexity. While it provides uniform analyses for many different phenomena, it also presupposes a rather complex interplay of three different types of features that need to be strictly ordered for the derivations to converge. Many important things happen in the numeration, which is seen as a pre-syntactic, generative component where features are assigned to and ordered on lexical items and syntactic heads, entire morphological paradigms are accessible for valuation via Agree, and lexical or syntactic items are kept both for first-time Merge and potential re-introduction as adjuncts later in the derivation. And, as noted above, not everyone will find the structure-removal approach appealing from a conceptual point of view, especially when it comes to ideas such as structural ‘abortion’, i.e. removing content that never even entered the derivation.

On another critical note, while this monograph is mostly very pedagogical and well written, it may annoy some readers that many of the footnotes in the book are very long; some even take up more than a page (e.g. fn. 59 in Chapter 1). Admittedly, many of these footnotes are very interesting, but their length and complexity do distract away from the main text.

These shortcomings are exceptions, though. All in all, this monograph is highly successful, both as a comprehensive introduction to German syntax and as a novel minimalist take on the empirical phenomena in question. It is characterised by a level of explicitness, honesty, and humility that is rare in the syntactic literature. It details each step of each analysis, mostly both writing these steps out and illustrating them with bracket notations or trees. It consistently links specific analyses to more general implications, predictions, and explanatory adequacy. Müller is honest about the limitations of his analyses, gives credit to competing approaches within different frameworks, and is careful to provide empirical support for his assumptions. At the end of Chapter 1, Müller humbly states that the monograph makes many decisions between several possible analyses which are not strictly empirically or conceptually supported in the book (p. 87). The fact that the book consistently draws attention to such implicit decisions rather than attempting to hide them is commendable, and in many ways, this is a textbook example of how a syntactic monograph should be written.

As its title indicates, this book will be of particular interest to linguists working on German, but there is every reason to expect that the analytical machinery applied here will be applicable to other languages as well. Müller frequently comments on cross-linguistic variation (or uniformity), most notably in connection with the passive in Chapter 3, and he bases his analyses on the existing literature on these syntactic phenomena in both German and other languages, especially English. This makes it an interesting read for anyone looking for state-of-the-art minimalist theory with an applicable toolkit.