3.1 The groundwork

Chapter 1 (‘Grammar’) is introductory. AR first gives a brief overview of a number of concepts used in traditional grammar (e.g. constituent, grammatical function, grammatical category, lexical/substantive category, functional category, noun, verb, finite forms, non-finite forms of verb, etc.). The section prepares the reader for the first discussion of the language data. AR then contrasts the taxonomic (i.e. ‘classificatory’) system of traditional grammar with the approach adopted by generative grammar, more specifically that adopted by Chomsky and those working within this framework, in which the focus is on the grammatical knowledge of the native speaker.

The goal of generative grammar is to offer a theory of Universal Grammar, i.e. to characterize what constitutes human language. In addition to rigid universal principles, UG also contains a set of parameters, choices that are set by individual languages on the basis of the empirical evidence. Parameters account for cross-linguistic variation. For instance, the so-called ‘pro-drop parameter’ (pp. 22–3) accounts for the difference between English (4a) and its Italian counterpart (4b): while in the former the subject pronoun must be overt, in the latter the subject pronoun can be omitted:

1. (4)

   1. (a) Maria thinks that *(she/he) speaks French.

   2. (b)

The pro-drop parameter will be set on the basis of evidence in the data: simplifying for expository reasons, the idea is that the child will come across sentences lacking an overt subject and will deduce that this option exists for his/her own language.

Chapter 2 (‘Structure’) introduces the concept of syntactic structure. On the basis of English data, AR elaborates the three-level clause structure (CP > TP > VP) that is by now standard in generative/minimalist approaches (see section 2). He introduces the internal structure of the phrase, based on the binary branching X-bar format: phrases are the result of binary merge operations. Importantly, after having discussed the format for structure, AR also provides the reader with a set of diagnostics to identify constituents. Most of these are the traditional diagnostics (pp. 58–60). The chapter also introduces the concept c-command, which is the core syntactic relation in generative syntax. In the final section of this chapter AR introduces Bare Phrase Structure (BPS) which replaces the older Xʹ-format. Because the traditional Xʹ-format is still widely used in generative work AR does not adopt BPS and also does not pursue its ramifications in the remainder of the book.

Chapter 3 (‘Null constituents’) shows in some detail that when analysing sentences we have to allow for null constituents, that is, constituents that are not given phonetic content, even though they are structurally represented. The chapter provides an interesting overview of a range of such null elements posited in the generative literature, with convincing argumentation for each of them. Starting from the intuitively fairly straightforward assumption that the Italian example (5a) has a null subject, it is shown that by the same reasoning imperatives (5b) and non-finite clauses such as (5c) can also be argued to have a null subject.

1. (5)

   1. (a)

   2. (b) Come back!

   3. (c) [To come back] would be a mistake.

Empirical evidence is provided for postulating a null subject in non-finite clauses; the null subject is represented as ‘PRO’ (roughly a ‘silent pronoun’).

1. (6) [_CP [_TP PRO To come back alone]] would be a mistake.

Pursuing the argumentation developed to postulate null subjects, AR extends the discussion and motivates postulating null auxiliaries, a null T(ense) in indicative clauses, subjunctive clauses and infinitive clauses, a null C in finite clauses and infinitive clauses, null determiners and quantifiers.

The chapter also focuses on the different types of infinitival to-clauses: it is shown that while some to-clauses have an overt (7a) or non-overt (7b) complementizer, other such clauses are ‘defective’ in that they lack the CP level (7c):

1. (7)

   1. (a) I would very much prefer [_CP for [_TP him to meet them]].

   2. (b) I would very much prefer [_CP Ø [_TP him to meet them]].

   3. (c) I believe [_TP him to be lying].

These latter sections prepare for the discussion of agreement and the feature content of T and C in chapter 7, and the discussion also prepares the ground for the discussion of phases in chapter 9.

3.2 Moving constituents

Chapter 4 (‘Head movement’) starts from the discussion of subject–auxiliary inversion in present-day English. The chapter shows that while present-day English does not allow V-to-T movement (8a), this movement was possible in earlier stages of the language (8b). In (8a) the lexical verb care cannot cross the marker of sentential negation; it remains adjacent to its complement.

1. (8)

   1. (a) *I care not about that.

   2. (b) I care not for her. (p. 153 (24a))

   3. (c) I don't care about that.

The final sections of the chapter focus on the syntax of English auxiliaries, including their interaction with negation, and with modal adverbs. The chapter also offers an analysis of do-support (8c).

At the theoretical level the chapter introduces the copy theory of movement, that is, the hypothesis that when a constituent is moved it leaves a copy in its base position, and the hypothesis that movement operations are restricted by locality conditions. So, for instance, in the Elizabethan English example (8b) the verb care, which originates in the position V, is moved to the position T, leaving a copy in V. We represent the copy by strikethrough:

1. (8)

   1. (d) I [_T care] not [_VP [_Vcare] for her].

The restrictions on locality introduced in section 4.7 of AR's book foreshadow the discussion of phase-based movement in chapter 9 and thus cleverly prepare the students for material that lies ahead.

Chapter 5 (‘Wh-movement’) widens the discussion of movement to that of maximal projections, focusing on movement to left peripheral position, as illustrated, for instance, by the movement of what languages in (9):

1. (9)

   1. (a) What languages can you speak? (p. 183 (1a))

In (9a) wh-movement encodes the scope of the question: (9a) asks for identification of the languages spoken. By analogy with the discussion in the preceding chapter it is proposed that movement of a wh-phrase (what languages in (9a)) leaves a copy in the base position.

1. (9)

   1. (b) What languages can you speak what languages?

Additional empirical support is provided for the copy theory of movement, which was introduced in chapter 4.

The empirical evidence provided comes from various sources: present-day English, earlier stages of the language and colloquial English. Especially the introduction of the latter type of empirical material is extremely valuable as it makes the students more aware of the linguistic phenomena that surround them. For instance, to show that a constituent that is moved from a lower clause into a matrix domain ‘stops over’ at the CP level of the lower domain AR gives (10), in which inside the tunnel is stranded in the CP of the lower clause.

1. (10) How far do you think how far inside the tunnel they went how far inside the tunnel? (p. 210 (67c))

The chapter also dwells on a number of technical issues related to wh-movement, namely so called pied-piping, i.e. the fact that in addition to a wh-word the containing constituent is moved to the left periphery (11a), pied-piping of a preposition (11b), long-distance wh-movement and the proposal that such movements apply cyclically (11c), multiple wh-questions (11d), and subject questions (11e).

1. (11)

   1. (a) Which assignment have you done?

      *Which have you done assignment?

   2. (b) They asked to whom he was referring.

      They asked whom he was referring to. (p. 203 (52))

   3. (c) Which picture of himself wasn't he sure that Mary liked best? (p. 213 (77))

   4. (d) Who might he think has done what? (p. 215 (83b))

   5. (e) Who on earth called the police? (p. 219 (92a))

In the final section of the chapter it is shown that the movement analysis proposed for the derivation of wh-questions extends to relative clauses and to exclamative clauses.

Chapter 6 (‘A-movement’) deals with the syntax of subjects and in particular with the movement of the subject from its merge position in the VP to the canonical subject position. The relevant movement is illustrated in (12): in (12a) the subject some students originates in the specifier of the VP, in (12b) it originates in the complement position of the unaccusative verb arise.

1. (12)

   1. (a) Some students should [_VPsome students get distinctions.]

   2. (b) Several complications have [_VP arisen several complications.]

The title of the chapter hints at the difference between the movement of the subject discussed here and leftward movement discussed in chapter 5 and instantiated by wh-movement – and other similar movements to the left periphery – illustrated in (13):

1. (13)

   1. (a) Which book do you think he will publish which book first?

   2. (b) When do you think he will publish this book when?

   3. (c) How do you think he will publish this book how?

While the fronted constituent which book in (13a) is an argument of the verb publish, this is not so for the fronted constituents in (13b, c). The specifier of CP is not an argument position because it may host both arguments (13a) and non-arguments (13b, c). Movement to the specifier of CP is movement to a non-argument (Aʹ) position, i.e. Aʹ-movement. On the other hand, in (12) the moved constituents are arguments.

A-movement can also cross sentential (TP) boundaries, as illustrated in (12c) and (12d):

1. (12)

   1. (c) Some students are likely [_TPsome students to [_VPsome students get a distinction]].

   2. (d) Some problems are likely [_TPsome problems to [_VP arise some problems]].

In both (12c) and (12d) the constituents some students and some problems are the subjects of the matrix verb be but they are thematically related to the lower verb phrase. In (12c) some students is the recipient of get a distinction, in (12d) some problems is the theme of unaccusative arise. It is proposed that the thematic relation between a verb and its arguments is established VP-internally. Thus the matrix subjects some students and some problems must have been first merged within the VP of the to-clause, and then they will be moved to the subject position of the matrix domain. The movement to the subject position is referred to as ‘A-movement’ because the subject position (SpecTP in the framework adopted) is generally reserved for arguments. Note, though, that the label A-movement is just a descriptive term, which is in fact descended from earlier versions of generative grammar and is not a primitive of the MP.

The chapter also contains a brief introduction to thematic theory (section 6.4), with as its core assumption the idea that lexical heads locally assign thematic roles and that thematic role assignment is uniform in that a particular thematic role is assigned to a specific position in the predicate. For instance, the agent thematic role is assigned to the specifier of the projection of the predicate.

The final section of the chapter compares raising patterns like those in (12c, d) and (14a) with control patterns introduced in chapter 3, in which an infinitival clause has a non-overt subject (PRO) whose interpretation is regulated by an argument – usually a subject – from the matrix clause (14b):

1. (14)

   1. (a) He does seem [to scare them]. (p. 266 (85a))

   2. (b) He does want [to scare them]. (p. 266 (85b))

Whereas in raising patterns (14a) the superficial subject (he) does not have a thematic relation in the matrix clause but rather is an argument of the embedded verb (scare), in the control pattern in (14b) the superficial subject he is thematically related to the matrix verb, i.e. want. According to the standard raising analysis for (14a), the subject of the matrix verb originates in the embedded clause and is moved to the canonical subject position in the matrix clause, and the copy of the subject in the infinitival clause is not spelled out, while in cases of control patterns as in (14b), the subject of the matrix clause is merged with the matrix verb and a null element is inserted as the subject of the infinitival clause.

1. (15)

   1. (a) He does seem [_TPhe to scare them].

   2. (b) He does want [_CP [_TP PRO to scare them]].

3.3 Developments in the MP

The final three chapters of the book become increasingly more technical and address some recent developments in the generative framework.

Chapter 7 (‘Agreement, Case and A-movement’) is concerned with the interactions between A-movement and agreement. The focus is on the role of features in the syntax.

AR examines the syntax of sentences such as those in (16).

1. (16)

   1. (a) It was said that he was taking bribes.

   2. (b) There were awarded several prizes.

It is proposed that the expletive pronouns it and there are merged in the specifier of VP and that they are attracted to the specifier of T by the EPP property of T (i.e. the requirement that T needs its specifier filled). AR makes the (not uncontroversial) assumption that the expletive has the feature [3 person] but unlike other nominals, the person feature here is uninterpretable (even though it has a value).

The chapter further examines the role of T and the associated EPP feature in control infinitives and in defective infinitival clauses, thus elaborating on two areas that were first introduced in chapter 3 and preparing the ground for chapter 9.

Chapter 8 (‘Split projections’) is concerned with an area of research that is often referred to as ‘cartography’. The core observation is that the three-level structure CP-TP-VP is too restrictive to accommodate the constituents of the clause and that enrichment by means of a more articulated structure is necessary. Articulated structures have been proposed for VP (Hale & Keyser Reference Hale, Keyser, Hale and Keyser1993, Reference Hale and Keyser2002; Pylkkänen Reference Pylkkänen2008), for the CP layer (Rizzi Reference Rizzi and Haegeman1997) and for the TP layer (Cinque Reference Cinque1999). Focusing on the interaction of so-called negative inversion (17a), and its interaction with wh-movement (17b), AR discusses the so-called split CP.

1. (17)

   1. (a) I am absolutely convinced that no other colleague would he turn to. (p. 325 (3))

   2. (b) Lee wonders whether under no circumstances at all would Robin volunteer (p. 327 (10b))

   3. (c) Syntax is the kind of subject which only very rarely will students enjoy. (p. 327 (9b))

The examples in (17) are all problematic for the unitary CP analysis since to the left of the subject we find more material than the unitary CP seems to allow for. For instance, in (17a), the subject he of the lower clause is preceded by conjunction that (in C), by a negative constituent no other colleague, and by an inverted auxiliary would. Thus it is clear that minimally to the left of the subject there must be an additional landing site for the auxiliary to move to. Similar arguments can be made for (17b, c).

Following (and adapting) Rizzi (Reference Rizzi and Haegeman1997), it is proposed that CP be decomposed into a hierarchically ordered set of projections, as in (18), each headed by a functional head associated with discourse related properties (‘illocutionary force’, ‘focus’, ‘topic’):

1. (18) ForceP > TopP > FocP > FinP

For (17a), for instance, it would then be proposed that that is in Force, that the fronted negative constituent no other colleague is in SpecFocP and the fronted auxiliary is in the head Foc:

1. (17)

   1. (d) [_ForceP that [_FocP no other colleague [_Foc would]. . .

The final sections of the chapter offer a detailed argumentation in favour of an articulated structure of the verb phrase, focusing on the role of the (abstract) ‘light v’ head, which is, among other things, associated with encoding the agentive thematic role. The chapter includes a discussion of transitive ergative structures (19a, b), unergatives, object control structures, unaccusative structures, passive sentences and raising structures.

1. (19)

   1. (a) They rolled the ball down the hill.

   2. (b) The ball rolled down the hill.

It is proposed that VP, the projection of the lexical head V, is merged as the complement of a ‘light (causative) v-head’, Ø, which is associated with the thematic role agent and which attracts the lexical V. Put very informally, (19a) is decomposed, as in ‘they caused the ball to roll down the hill’. In the VP shell, the object DP the ball is merged as the specifier of VP, the PP down the hill is the complement. The proposal would be in line with the idea that there is a ‘subject–predicate’ relation between the ball and down the hill: as a result of this rolling, one could say that the ball went down the hill. At the level of causative v, the agentive subject they is merged as the specifier of the abstract ‘light’ causative v. V roll moves to v. (19a) has the representation in (19c). (19b) lacks the agentive interpretation and has the structure in (19d), where the light v has no external argument:

1. (19)

   1. (c) [_{v P} they [_v roll Ø] [_VP the ball [_Vroll] [_PP down the hill]]]

   2. (d) [_{v P} [_v roll Ø] [_VP the ball [_Vroll] [_PP down the hill]]]

The decomposition of the VP in terms of a layered structure is often referred to as the ‘VP shell’ analysis (see Larson Reference Larson1988 for the first proposal). We return to these structures in section 4.

Chapter 9 (‘Phases’) offers a very dense introduction to recent developments in minimalist research on the concept ‘phase’. The core ideas of phase theory are introduced, i.e. the proposal that the computational component of the language faculty does not build the structure of a clause as a whole, but rather that the structure is built up cyclically in chunks, or ‘phases’ (Uriagereka Reference Uriagereka, Epstein and Hornstein1999; Chomsky Reference Chomsky, Martin, Michaels and Uriagereka2000 et seq.). Once a phase is formed, it is transferred to the semantic (LF) and phonological (PF) component (see (3)) and becomes unavailable for any further syntactic operations. Constituents which are internal to a phase and which need to be moved to external areas must therefore be evacuated before the point of Transfer. This is done via the periphery of the phase or, to use the technical term, the ‘edge’ of the phase. The edge features (EF) on the phase head allow a constituent to move from a phase-internal position to the phase edge (see Chomsky Reference Chomsky, Gärtner and Sauerland2007, Reference Chomsky, Freidin, Otero and Zubizaretta2008), which will allow a constituent internal to the phase to evacuate via the specifier of the phase head. There is a consensus that phase heads are C and the causative light v (often represented as v* in the literature).

The discussion in chapter 9 brings together components of all the preceding chapters, including the issue of what constitutes the trigger for wh-movement, a point that was left rather open-ended in earlier chapters, as well as the syntax of defective clauses, a domain that had been fairly prominent in a number of preceding sections. In this sense it is an appropriate final chapter for the book.

Though we return to a more general evaluation of the book in later sections, we wish to point out here that chapter 9 is extremely dense and highly technical. In it, AR introduces assumptions that are fairly standard in Minimalism, as well as many new avenues which are not (yet) generally adopted in the literature.

4.1 Empirical coverage

It is clear from the survey of the chapters in section 3 that the empirical range of the book is quite broad: the main components of the English clause are covered in the discussion, and for some aspects of the syntax of English sentences the reader gets quite a detailed picture. In this respect, to our minds the most successful sections are chapter 3 on null constituents and sections 8.6–8.10 on the decomposition of the VP. Both of these bring together and highlight data that are not usually treated together in such a systematic way in traditional approaches to English grammar. In both chapters, the argumentation is very convincing, with the repeated use of the same type of argument extended to new patterns. Let us illustrate this for the decomposition of the VP.

The discussion of the internal structure of the verb phrase starts out from so called ‘transitive ergative structures’ (p. 345) illustrated in (20).

1. (20)

   1. (a) They will roll the ball down the hill. (p. 347 (67a))

   2. (b) The ball will roll down the hill. (p. 347 (67b))

As discussed in section 3, based on the pairing of the sentences in (20), the traditional VP is decomposed into two separate ‘shells’: the (lower) VP shell encodes the relation between the ball and down the hill as mediated by the verb roll in (20b). This VP-shell is then embedded under a higher abstract causative ‘light’ verb represented as v, whose subject argument is they. The light verb is affixal in nature: the lexical V roll is inserted in V and moves to adjoin to the abstract affixal v.

1. (20)

   1. (c)

>In the discussion AR argues that, given this analysis, and under the assumption that adverbs are adjoined to the intermediate level of representation – an assumption which, however, is not without its problems – the fact that an adverb such as gently may either precede roll or follow the ball and precede down the hill is unproblematic. In (20d) gently is adjoined to vʹ, in (20e) it is adjoined to Vʹ.

1. (20)

   1. (d) They gently rolled the ball down the hill.

   2. (e) They rolled the ball gently down the hill.

The decomposition of VP is extended to other patterns, illustrated in (21). We provide a partial bracketing.

1. (21)

   1. (a) they will [_{v P} (carefully) load [_VP the truck (carefully) with hay]]. (p. 352 (85))

   2. (b) He was [_{v P} taking [_VP the rubbish out]]. (p. 353 (87))

   3. (c) What [_{v P} (eventually) decided [_VP you (eventually) to take syntax]] (p. 357 (103))

   4. (d) The horse was [_{v P} jumped [_VP (perfectly) over the fence]] (p. 365 (126))

For some of these examples, the proposed analysis is again supported by means of adverb placement, and what is attractive about this is that the same argumentation is applied systematically to a range of patterns. However, it is a pity that the argumentation is not systematically applied to all examples. For instance, while for the verb load the two adverb positions are illustrated (21a), this is not so for the verb-particle combination take out in (21b), where the variants in (22) are not provided. Similarly, for (21d) AR provides the example with the postverbal adverb but an example along the lines of (23), with preverbal adverb, is also not given. Observe, though, that extending the argumentation to these examples is not obvious. Indeed, the native speakers we consulted found the adverbials in the positions indicated at least marked if not downright unacceptable.

1. (22)

   1. (a) He will eventually/quietly take out the rubbish.

   2. (b) He will take the rubbish *eventually/?/*quietly out.

2. (23) The horse was ?/*eventually/*perfectly jumped over the fence.

AR extends the layered VP to ‘ordinary’ transitive sentences such as (24a), with an agentive subject, to which he assigns the structure in (24b):

1. (24)

   1. (a) He read the book.

   2. (b)

Now at this point we suspect that the reader may be puzzled. The representation in (24b) (AR's (91) on p. 354) is prefaced by the following passage: ‘Chomsky (Reference Chomsky1995 [no page reference]) proposes a light-verb analysis of two-place transitive predicates.’ However, it seems to us that (24b), which is not motivated, will raise questions in the light of the preceding argumentation, as we will show in what follows.

For resultative type constructions such as (25a) a layered VP structure as in (25b) is proposed in which the ‘direct object’ occupies the specifier of the VP shell, and the subject originates in the specifier of the vP-shell.

1. (25)

   1. (a) The acid will turn the litmus paper red.

   2. (b) The acid will [_{v P}the acid turn [_VP the litmus paper turn red]]. (p. 354 (89))

Since (26a) is listed along with (25a) (p. 353 (88)), we assume it should be assigned the structure in (26b):

1. (26)

   1. (a) They may paint the house pink (p. 353 (88b))

   2. (b) They may [_{v P}they paint [_VP the house paint pink]].

It seems to us that following the argumentation in the chapter, the reader will definitely be tempted to then analyse an example with transitive paint such as (27a) as (27b), since it would appear that the DP the house has the same relation to paint in (27b) as in (26b). Possibly, inspired by the argumentation in chapter 3, the student may even postulate a null predicate in the lower VP shell as in (27c):

1. (27)

   1. (a) They may paint the house.

   2. (b) They may [_{v P}they paint+Ø [_VP the house paint]].

   3. (c) They may [_{v P}they paint [_VP the house paint Ø]].

Analogously, the student might propose that (28a) has the structure in (28b), where widen is interpreted as ‘cause to widen’, ‘cause to become wider’. Once again one might conceive of widen itself as being decomposed into wide and en, with wide originating as the predicate in the lower VP shell (28c), with wide being attracted by the affix en:

1. (28)

   1. (a) They may widen the road.

   2. (b) They may [_{v P}they widen+Ø [_VP the road widen]].

   3. (c) _ [_VP the road [_V -en] [wide]]].

And by analogy with such transitive examples, the student might then adopt (29a) for all transitive verbs and propose (29b) instead of AR's unmotivated (24b) for the simple transitive sentence.

1. (29)

   1. (a) [_vPsubject V+Ø [_VP DP V]].

   2. (b) He [_{v P}he read [_VP the book read]].

Adverb placement, the empirical argument advocated for other representations, does not help in choosing between the two patterns: the preverbal adverb in (30a) could be adjoined to vʹ, and the postverbal adverb in (30b) could be adjoined to Vʹ.

1. (30)

   1. (a) He carefully read the book.

   2. (b) He read the book carefully.

   3. (c) He [_vPhe ([_vʹ carefully) read [_VP the book ([_Vʹ carefully) read] (])]].

Indeed it is not obvious where the adjunct in (30b) would adjoin in (24b). One might think of right adjunction to vʹ (30d) or to VP (30e), for instance, but note that since Kayne (Reference Kayne1994) – see also section 6.1 – right-hand adjunction is rather frowned upon.

1. (30)

   1. (d) He [_vPhe [vʹ read [_VP [_VPread the book]] carefully]]

   2. (e) He [_vPhe read [_VP [_VPread the book] carefully]]].

It seems to us that inquisitive students will inevitably come up with these questions, and that they should be encouraged to do so, but the book does not give the instructor a way of handling these and it is disappointing that the pattern of argumentation that is developed in the book is not sustained across the full paradigm. Rather, disappointingly, the only argument for (24b) seems to be recourse to ‘the voice of authority’. Of course, the instructor might go to the references section and read the various works cited, but this will be a time-consuming job and it would be helpful if the textbook itself provided more help here.

4.2 English sentences

As we have observed, in some sections of the book the argumentation is excellent and instructors will be able to use the material presented not just for teaching about English, but also for teaching the methodology of formal syntax.

However, there are also areas where consistent application of the argumentation provided leads much further. One case is illustrated in the discussion of non-finite clauses in chapter 3, section 10. In chapter 2 the traditional diagnostics for structure are introduced, one of which is ‘coordination’. AR (p. 59) proposes the coordination condition in (31):

1. (31) Co-ordination condition (p. 59 (50))

   Only constituents of the same type can be coordinated.

The application of the condition is illustrated on p. 59 with examples such as (32a). In a long parenthetical section on p. 60, AR briefly addresses problematic cases in which the constituents are not at first sight of the same type (32b).Footnote ³

1. (32)

   1. (a) The chairman has resigned from the board and the company. (p. 59 (51a))

   2. (b) He's cross with her and in a filthy mood. (p. 60)

In later sections, though, the potential problem raised by the existence of examples such as (32b) for the application of the coordination condition is not taken into consideration and the condition is implemented in its strongest version. Hence, from the fact that two constituents, which at first glance seem similar in structure, cannot be coordinated it is concluded that they are not of the same type. On the basis of the fact that the bracketed embedded clauses in (33a) and (33b) cannot be coordinated (33c), it is concluded that they must be of different types: one is a projection of TP (33d), the other is a full CP projection (33e):

1. (33)

   1. (a) We didn't intend [you to hurt anyone].

   2. (b) We didn't intend [for him to hurt you].

   3. (c) *We didn't intend [you to hurt him] or [for him to hurt you]. (p. 127 (101))

   4. (d) We didn't intend [_TP you to hurt anyone]. (p. 127 (101))

   5. (e) We didn't intend [_CP for [_TP him to hurt you]].

We note in passing that the judgment in (33c) is not uncontroversial. One of our informants in fact accepted this example.

Conversely, when two apparently dissimilar strings can be coordinated this becomes the basis for concluding that they are of the same type, again with the strict interpretation of categorical identity. Thus, the fact that an infinitival clause introduced by for (34a) and a control clause (34b) can be coordinated (34c) is taken as evidence that both must be CP, and that the control clause instantiates a null complementizer (34d):

1. (34)

   1. (a) I will arrange [for my wife to see a specialist].

   2. (b) I will arrange [to see a specialist]

   3. (c) I will arrange [to see a specialist] and [for my wife to see one at the same time].

   4. (d) I will arrange [_CP Ø_for [_TP PRO to see a specialist]].

The difference between to-infinitival clauses that are mere TP and those that are dominated by a CP with null C is reflected in the possibilities of passivizing the subject of the infinitival clause: bare infinitival TPs allow passivization (35b), while infinitival CPs do not (36b).

1. (35)

   1. (a) Most members of the council considered these proposals to be unacceptable.

   2. (b) These proposals were considered to be unacceptable by most members.

2. (36)

   1. (a) Most members of the council preferred the discussion to be postponed.

   2. (b) *The discussion was preferred to be postponed by most members.

The student who applies the coordination test illustrated in (33) may face some unexpected results. Consider, for instance, the following attested examples:

1. (37)

   1. (a) It's been a total witch-hunt. People just want it over and for him to carry on dealing with the financial side of things. (The Guardian, 13/2/1999, p. 3, col. 8)

   2. (b) Blatter wants doping cases assessed individually, rather than automatically imposing a blanket ban, and for this to be part of the agreement with Wada. (The Guardian, 18/5/2004, p. 20, col. 8)

In (37a) and in (37b) a for clause is coordinated with what could be described as a tenseless clause (Quirk et al. Reference Quirk, Greenbaum, Leech and Svartvik1985, Huddleston & Pullum et al. Reference Huddleston and Pullum2002) or a small clause (Aarts Reference Aarts1992), that is: a proposition-type complement containing a subject–predicate relation which is not mediated by a copular element. Given that in both examples above these (underlined) ‘small clauses’ are coordinated with for clauses, it would then follow that they are CPs. On the basis of the coordination test, the conclusion (at first sight surprising) must be that at least some so-called small clauses are not so ‘small’ after all and are to be aligned to non-finite CP complements (for similar discussion see also Kayne Reference Kayne1984: 37 and n. 33; Coopmans & Stevenson Reference Coopmans and Stevenson1991; Starke Reference Starke1994; Citko Reference Citko2008; Belletti Reference Belletti2009: ch. 10). However, the passivization test confirms that small clauses and CP complements must be distinguished: the subject of the small clause complement of consider can be passivized, while that of the small clause complement of prefer cannot.

1. (38)

   1. (a) Most members consider these proposals unacceptable.

   2. (b) These proposals are considered unacceptable.

2. (39)

   1. (a) Most members preferred [these proposals left out of the discussion].

   2. (b) *These proposals were preferred [left out of the discussion].

In the book English small clauses only get the briefest of mentions (pp. 111, 447). This is regrettable since, though perhaps currently less central to the theory (see Alhorais Reference Alhorais2007; Citko Reference Citko2008), the empirical domain of small clauses is definitely one that received a lot of attention in earlier stages of the generative framework and new insights have been gained (see Stowell Reference Stowell1983; Starke Reference Starke1994).

On the positive side, we wish to point out that precisely because AR provides the tools for the analysis, and because he includes a wide array of data from English, we think that in a teaching context, and in particular in one in which the focus is on English sentences and in which the audience also comprises native speakers, issues such as the one outlined above will quite naturally come up. As such, this is a very positive feature of the book because it will encourage independent exploration of linguistic data.

As already pointed out in relation to example (33c), the judgments put forward in this book are not always uncontroversial and strong theoretical conclusions are sometimes drawn from data that are controversial. This is clearest in chapter 9 in which AR elaborates the theory of feature inheritance on the basis of controversial judgments. For reasons of space we cannot go into this issue here but we refer the interested reader to Grewendorf & Kremers (Reference Grewendorf and Kremers2009).

4.3 Universal grammar and the use of comparative material

A very attractive feature of the book is that AR does not usually (but see below for some additional comments) limit the discussion to a restricted set of sentences drawn from the technical literature. He offers a rich array of empirical data from standard English (see for this especially the data discussed in the discussion of the split CP) and he also discusses data drawn from language acquisition, from diachronic sources (where Elizabethan English features prominently), from colloquial English, from Hiberno-English and even from performance errors such as slips of the tongue, for which he shows convincingly that they may be the reflex of underlying syntactic principles. Anyone teaching English syntax can only be overjoyed by this wide approach to the data.

However, on the reverse side, it is disappointing that the focus is so much on English and that discussion of other languages is so restricted, and we especially regret that the care that is given to the English data is not extended when it comes to comparative discussion. For instance, in the discussion of the prodrop phenomena on p. 92 the author focuses on the non-overt subject in (40). Observe that in the discussion in the book there are no specific glosses for the morphology, so our (5) was actually given as (40), with pro representing the understood subject:

1. (40) Si, pro è tornata.

   yes, pro is returned

   ‘Yes, she has returned.’

AR says that one argument for positing pro is that ‘it relates to the agreement morphology carried by the auxiliary è “is” and the participle tornata “returned” in [40]. Just as the form of the (third-person singular) auxiliary è “is” and the (feminine singular) participle tornata is determined via agreement with the overt (third-person feminine singular) subject Maria . . . so too the auxiliary and participle agree in exactly the same way with the null pro subject in [40]’ (p. 92). Presented with (40), a reader who does not know any Italian will not be able to see that what makes Italian different from, say, French or English is that the agreement paradigm is much richer in the former with a specific inflectional form for each person and number combination. What worries us about the approach in AR's book is that the data provided do not allow the student to see for him/herself that agreement is indeed rich in Italian and that the participle also shows agreement. Rather, they simply have to accept what they are told.

In other, according to our point of view even worse, cases the data are not even provided. For instance, on p. 146, AR writes: ‘Such an interrogative affix analysis is far from implausible from a cross-linguistic point of view, in that (e.g.) yes–no questions in Latin could be formed using the overt question suffix -ne, and this could attract a finite verb to attach to it.’ No data are provided at all. Similarly, on p. 202 in the discussion of the syntax of the genitive whose in whose car we find: ‘and indeed there are a number of languages which have a type of possessive structure paraphrasable in English as whose the car – e.g. Hungarian’. In other cases, we think the discussion of comparative material is potentially misleading. For instance, on p. 165 we find: ‘just as in the ne. . .pas “not . . . at all” negative in present-day French, ne has dropped out of use in colloquial styles’. Clearly in present-day French pas is not equivalent to at all, which conveys some special emphasis on the negation, and again a reader who has no knowledge of French will easily be misled by this throwaway remark.

6.1 Missing components

Although AR covers a wide range of topics, some important ones go unmentioned. One striking omission is that hardly any attention is being paid to the level of Logical Form (LF). This level is, for instance, invoked to encode the interpretation of multiple wh-questions such as (42a) which have a paired reading of the wh-constituents, as suggested in the answers in (42b):

1. (42)

   1. (a) Who will buy what?

   2. (b) Mary will buy the wine and Bill will buy the bread.

It is true that LF is less prominent in the recent literature, but it has played a crucial role in the development of the framework, and apart from being mentioned as one level of representation (pp. 192, 465, 478) it is not discussed any further. Other areas that might have received attention are Kayne's (Reference Kayne1994) Linear Correspondence Axiom (LCA), Cinque's (Reference Cinque1999) work on adverbials and alternative theories of control, the latter already pointed out in section 5.

Kayne's theory (Reference Kayne1994) is a theory of linearization and his idea is that hierarchy maps onto precedence through the LCA. His work is crucial for the discussion of word-order variation. Kayne proposes that the variation between VO order and OV order is not one of parametric variation in the head-complement order of VP but that the VO order is the default order and that OV order is derived by leftward movement of the complement. Kayne's work has been very influential and advanced students of syntax need to know at least the basics of the LCA. It is crucial in order to be able to read much of the literature from 1994 until today. It is puzzling that AR does not discuss the LCA. Especially because the last two chapters of the book aim at getting the student up to date on the most recent theoretical developments. Compared to the intricacies of phase theory laid out in chapter 9, one would think that the LCA is pretty straightforward to spell out. The decision to leave aside the highly influential LCA is unexpected and we think the book would have benefited from including a section on the LCA.

In the same chapter, though, the sections on the split TP, for which the projections AspP and MoodP are proposed, seem redundant: as acknowledged by AR himself, the phenomena discussed in the section on AspP are partly covered by the sections on vP/VP in the same chapter, and the section on MoodP is shown to be problematic. In this context, then, it is surprising that there is no more extensive discussion of Cinque's (Reference Cinque1999) hypothesis concerning the articulated TP. AR does refer to Cinque's work, but he does not address his theory in detail. We think this is unfortunate since Cinque's work is widely known and has led to a lot of discussion in the field (see, for example, Ernst Reference Ernst2002, Reference Ernst2007, Reference Ernst2009; Nilsen Reference Nilsen2004). Rizzi's work on the split CP (discussed by AR in section 8.2) and Cinque's work on the articulated TP – not discussed here – have been very important for the development of ‘cartographic syntax’ (see above), and the work also raises questions for those wishing to rigidly stick to a minimalist functional structure. As such, it seems to us a strange decision not to discuss Cinque's contribution to the debate and instead discuss less influential work

6.2 The historical context

Since the MP is a programme, it is hard to identify its stable parts as it is in continuous development. In this book the MP is presented almost in a vacuum. It is not shown how the MP has evolved from earlier formats of the generative model. As we have argued above, the MP is embedded in the earlier generative framework. In order to identify its core components and to understand their developments, it is important to understand the foundations on which the MP rests. That is, one has to know the core parts of GB and what the differences between earlier incarnations of generative syntax (such as GB) and the MP are. Although scholars may have different opinions as to how important the GB-framework is for understanding the MP, we think that some historical perspective – for example, of the sort that one finds in Hornstein, Nunes & Grohmann (Reference Hornstein, Nunes and Grohmann2005) – can contribute towards a better understanding of why the theory has evolved the way it has. That will enable the students to get a good sense of the big picture of how the thinking about the language faculty has developed in the generative framework. AR does not say much, if anything, about earlier models in his books, which makes it harder for the students to understand how and why the MP developed the way it did.