In the course of the Modern English period, English has undergone a massive restructuring of its system of clausal complementation. Despite the existence of a growing number of corpus-based studies dealing with relevant issues, many aspects and domains of what has been referred to as the Great Complement Shift (cf. e.g. Vosberg Reference Vosberg2006; Rohdenburg Reference Rohdenburg, Dalton-Puffer, Kastovsky, Ritt and Schendl2006) still remain unexplored. Tackling two kinds of complement shift, the decline of that-clauses and their (partial) replacement by to-infinitives as well as the later expansion of various gerunds at the expense of to-infinitives, this book promises to be an important addition to the field of historical English syntax. The book is divided into six chapters including the introduction and a final one providing a ‘summary and conclusions’. The bulk of the material is based on articles previously published in various edited books and journals.

Chapter 1 (pp. 2–25) introduces us to the eleven verbs selected for this study (forbid, refuse, forbear, avoid, prohibit, prevent, hinder, refrain, fear, doubt, deny), outlining the semantic and syntactic features that unite them. Briefly, all of these verbs are found to have been associated in Middle English and/or Early Modern English with two syntactic phenomena, (a) the use of expletive negation and (b) the (partial) replacement of that-clauses by but-clauses. Accordingly, the occurrence of these features is duly noted throughout the book. However, Iyeiri does not provide a typological underpinning for the use of expletive negation, as is done by several other authors. Many of the phenomena and individual verbs dealt with by Iyeiri have been discussed before in relevant studies, which are not always acknowledged. While Fanego's work (e.g. Fanego Reference Fanego1996) and Vosberg (Reference Vosberg2006) are referred to occasionally, the analysis of the twelve ‘negative verbs’ in Rudanko (Reference Rudanko2000: 109–45) is not mentioned at all. Chapter 1 contains a non-committal discussion of some real and claimed contrasts between infinitival and gerundial uses. In this connection, it would have been helpful to refer as well to Kjellmer (Reference Kjellmer, Allwood and Ljung1980), who shows that the choice of the gerund clearly correlates with a given verb's affinity for nominal complements. The chapter concludes by presenting the main database, the quotations contained in OED2, and a number of minor ones, which are only drawn upon less systematically.

Chapter 2 (pp. 27–49) focuses on the two verbs forbid and refuse, which have been mainly associated with the to-infinitive over the last few centuries. Concerning forbid, Iyeiri shows that the occurrence of that-clauses precedes that of infinitives and gerunds (in that order) and that gerunds containing from represent an even more advanced stage of development. However, an important syntactic distinction, the presence or absence of directly linked personal objects preceding the clausal complement (whether finite or non-finite), fails to be addressed on a systematic basis in the text or the figures supplying the statistical evidence per century. The failure is all the more surprising since the illustrative material on pages 28–42 exemplifies all of the historical possibilities in the active shown in (1)-(3).

1. (1) (a)/(b) forbid (someone) (that) NP should do this

2. (2) (a)/(b) forbid (someone) to do this

3. (3) (a)/(b) forbid (someone) (from) doing this

The distinction is important for three reasons:

1. 1. In the default case, the relevant replacement processes (exchanging finite clauses for infinitives and infinitives for gerunds) could be expected to preserve the existing situation either with or without preceding NP.

2. 2. It is a fact that with directive verbs (amongst others) the tendency to phase out finite clauses affected the more complex cases containing additional nominal complements earlier and more drastically than the simpler ones without them (Rohdenburg Reference Rohdenburg1995). With infinitives, by contrast, the earlier potential absence of a controller NP has been giving way to its obligatory presence with almost all classes of verbs.

3. 3. Finally, while the from-less gerund is still found after manipulative verbs without a preceding NP, the more explicit from-containing gerund has virtually always been dependent on the presence of an explicit object controller. Exceptions to this tendency are provided by reflexive uses such as keep (oneself) from -ing.

The inclusion of an extraposed that-clause after It is forbidden (p. 33) suggests that Iyeiri does not distinguish between actives and passives, either. I will return to this issue later on.

I am also baffled by Iyeiri's refusal ‘to compare the raw frequency of infinitives in the fourteenth century with that in the fifteenth century and say that the use of infinitives declined during the period’ (p. 32). On the basis of the calculations supplied by Hoffmann (Reference Hoffmann2004), who is quoted by Iyeiri on at least two occasions, it should have been possible to work out normalised frequencies allowing us to carry out such comparisons.

While generally speaking types (1a-b) may have been phased out by the present time, there is a fossilised use involving the string God forbid, which is still acceptable today. Iyeiri shows that this special use possesses two distinctive properties: (a) the non-existence of expletive negation and (b) the absence of a preceding object NP.

Compared with forbid, the history of refuse is relatively uneventful. According to the OED, the verb has never governed that-clauses, and since its first attestation in Late Middle English it has typically been associated with to-infinitives. The use of gerunds, though attested since the eighteenth century, has not gained a firm foothold with this verb, either. Unfortunately, no serious attempt is made to demarcate precisely the class of clausal gerunds replacing to-infinitives. In footnote 30 (p. 32), the policy Iyeiri has adopted is described as follows:

Moreover, another important distinction fails to be drawn here or elsewhere: since the implied subjects of the infinitives after refuse are virtually always coreferential with the subjects of refuse, the gerunds replacing them should be similarly restricted, excluding, for example, any uses involving arbitrary control. Finally, Iyeiri points out an example of what used to be described as a subject-to-object raising structure (p. 46). Again, it is not clear whether such phenomena are included in the statistics of the infinitive category.

Chapter 3 (pp. 52–77) deals with forbear and avoid, which in addition to infinitives experienced a considerable expansion of the directly linked gerund during the Early Modern English period. In the case of forbear, the rise of the infinitive is assumed to have been preceded by a stage characterised by that-clauses (there are only three examples in the OED between the thirteenth and fifteenth centuries), and by the eighteenth century the infinitive is clearly outnumbered by the upcoming gerund, especially after the strings can/could not forbear. However, over the last two centuries, gerunds have declined again (at the expense of infinitives), with the from-containing variant becoming at least as frequent as the simple one.

In contrast to the history of forbear, that of avoid is relatively straightforward. The gerund was more frequent than the to-infinitive as far back as the sixteenth century, and by the eighteenth century the infinitive was virtually phased out. However, it is an open question whether the that-clause can be regarded as the direct precursor of the infinitive or the gerund. The only that-clause (of perhaps two attested in the OED) that Iyeiri quotes for the sixteenth and seventeenth centuries happens to contain a subject that is not coreferential with the (implied) subject of the superordinate string to avoid (p. 67). In this connection, it is unfortunate that Iyeiri refrains from giving absolute frequencies of the three rivalling broad categories: I was unable to retrieve any that-clauses complementing avoid for the seventeenth century (listed as occurring in 3 per cent of the rivalling complement patterns in figure 9 on p. 200).

Unlike Vosberg (Reference Vosberg2006) and a few other authors, Iyeiri does not usually pay much attention to well-established contextual constraints such as horror aequi, complexity phenomena including structural discontinuity and the use of extractions. It is only with respect to the verb avoid that she explores the role of horror aequi in some detail.

Incidentally, some of the examples illustrating the category of gerunds rivalling to-infinitives appear to be ill-chosen. Consider, for instance:

1. (4) c1590 To avoid ensuing jars Ile hamper vp the match, Ile . . . wed you here. (p. 74)

2. (5) 1617 The Curtizans make all the forepart of their gownes in like manner open, to avoid wrinckling. (p. 74)

In (4), ensuing clearly functions as an attributive modifier of the noun jar, and in (5), wrinckling most likely represents a non-clausal complement rather than a subject-controlled gerund dependent on to avoid.

Chapter 4 (pp. 79–117) is devoted to the verbs prohibit, prevent, hinder and refrain. Concerning prevent and hinder, the three patterns illustrated in (6a-c) are contrasted in great detail, and it is shown that the decline of the possessive/genitive alternative in (6a) occurs at a slower rate with personal pronouns than with the set of all other NPs.

1. (6)

   1. (a) prevent his/the man's leaving

   2. (b) prevent him/the man leaving

   3. (c) prevent him/the man from leaving

The major finding covering all four verbs is that complex gerunds containing from have largely replaced their prepositionless counterparts in the course of the Late Modern English period. The fact that in the case of prevent (and some other verbs not treated by Iyeiri) British English has been undergoing a reversal of this change in the late twentieth century (see e.g. Mair Reference Mair2002: 112–15) does not appear to be reflected in the OED quotations and is only mentioned in passing by the author.

Unfortunately, the discussion is marred again by the failure to draw important distinctions between different types of finite, infinitival and gerundial constructions. Thus it is not clear to me whether a non-clausal complement like (7) is included in the gerund category.

1. (7) Canada prohibits the broadcasting of ‘spot ads’. (p. 86)

Moreover, the assumption (which I share) that with prohibit the prepositionless type (6b) had been expanded in the Late Modern English period to give type (6c) including from does not square with examples like (8) involving arbitrary control, where the addition of from would require a preceding controller object.

1. (8) . . . the Directive on the Conservation of Wild Birds which prohibits keeping and trading in wild birds . . . (p. 87)

Perhaps the most important factor determining the use of from or zero with gerunds governed by prevent (and several other verbs) in present-day British English is the voice contrast, with passives being virtually always confined to the more complex and explicit variant (see e.g. Andersson Reference Andersson1985: 101–3). Having analysed the OED data relating to prevent for the time span 1700–1987 myself (Rohdenburg Reference Rohdenburg, Plag and Schneider2000: 36–7), I was hoping to learn a bit more about the evolution of this asymmetry, but regrettably the topic is not even touched upon.

Chapter 5 (pp. 119–75) is concerned with the verbs fear, doubt and deny, where infinitival and gerundial complements continue to this day to be far less common than that-clauses. Even so, the that-clause is also argued to have been in decline during part of the Modern English period, at least with fear and doubt. The argument is linked to the rise of ‘epistemic parentheticals’, which is assumed to be reflected in the following developments:

• – the increasing use of that-clauses with ‘that unexpressed’ after the string first person + present tense verb;

• – the increasing use of ‘dislocated’ uses such as (9).

1. (9) 1598 Take my armour quickly, 'twill make him swoune, I feare. (p. 131)

Iyeiri refrains from quantifying the ‘epistemic’ uses of the strings I/we fear and I/we (do not) doubt any more precisely, and in view of the mass of context-reduced quotations to be processed I can sympathise with her decision. But perhaps one of the tests suggested by the pragmatic concept of dominance might have been helpful in this respect?

In the case of that-clauses, Iyeiri does not distinguish between subjects that are coreferential with the subject of the superordinate verb and those that are non-coreferential. This is an important distinction for two reasons:

• – Most infinitives and gerunds of at least two verbs, fear and deny, display subject control, suggesting that they should be directly compared with finite complements involving coreferential subjects.

• – In line with the Complexity Principle (see e.g. Rohdenburg Reference Rohdenburg1995, Reference Rohdenburg, Plag and Schneider2000, Reference Rohdenburg, Dalton-Puffer, Kastovsky, Ritt and Schendl2006), it has been claimed by several researchers that the omission of that is favoured with coreferential subjects (because they are easier to process; see e.g. Elsness Reference Elsness1984).

Finally, Iyeiri observes that that-clauses containing that have been on the increase again in the twentieth century with all three verbs. In this connection, the possibility needs to be checked as to whether and how far this finding is simply due to an increase of non-coreferential subjects in the OED quotations.

The book is a bit too repetitive for my liking. Chapter 6 (pp. 177–98) provides a lengthy summary of the findings presented in chapters 2–5, which had already been summarised extensively in the introduction and in various parts of chapters 2–5.

In conclusion, a great deal of hard work has gone into this book, which contains a wealth of data presented in several tables and thirty-eight figures. This certainly demonstrates the usefulness (and occasionally the limitations) of the OED as a tool for analysing linguistic change and variation. No doubt, it is very useful to have in one place a collection of in-depth analyses covering at least the Early Modern and the Late Modern English periods. On this basis alone, the book would be essential reading for anybody interested in the field of English historical linguistics or simply the main database, the OED quotations. It is a pity, therefore, that overall the book does not achieve the degree of theoretical sophistication and care displayed by researchers such as Fanego, Rudanko, Mair, Vosberg or De Smet & Cuyckens (Reference De Smet, Cuyckens, Cain and Russom2007) in relevant contributions. Moreover, the conceptual and terminological laissez-faire attitude prevailing in much of the book makes it difficult to compare like with like in the replacement processes involved or to check the statistical data supplied. Finally and regrettably, there are a number of stylistic infelicities, which may be largely attributed to poor copy-editing.