This book is a welcome addition to the linguistic literature on modality, where research has largely concentrated on modal verbs, and studies of other form classes, such as adjectives, have been relatively few. Van linden's book fills this gap in the literature by presenting a detailed study of modal adjectives (e.g. essential, crucial, important) in synchrony and diachrony. As Van linden not only discusses the meaning and semantic change of modal adjectives, but also their clausal complementation patterns, this book will certainly be of interest to a wide audience, including those interested in historical linguistics, semantics and syntax.
The book starts with an introduction, which provides a concise summary of the main topics of the book and includes some fundamental terminological and theoretical distinctions. Based on her data on English modal adjectives, Van linden proposes a ‘conceptual map’, which she argues to have both synchronic and diachronic validity. In her conceptual map, she distinguishes between dynamic modality, deontic modality and non-modal evaluation. Dynamic meanings relate to necessities which are internal to the States of Affairs (SoA), and which therefore do not express the attitude of a modal source. For example, in (1), the necessity expressed by the modal adjective essential is independent of the attitude or the desire of the speaker.
(1) Air is essential for humans to survive.
Deontic meanings, by contrast, are attitudinal assessments of potential SoAs, as in (2), where the speaker is the modal source assessing the necessity of the distribution of the money.
(2) It was essential, he said, that money was better distributed, so that it reached the poorest people.
Contrary to dynamic and deontic expressions, where the complement clause expresses a potential SoA, the complements of non-modal evaluations express information that is presupposed to be true, as in (3).
(3) As an IT consultancy, it's appropriate we're taking the initiative and using the latest IT technology.
As will become evident in the later chapters of the book, these three categories are relevant both to the semantic development of modal adjectives and to their patterns of polysemy.
Chapter 1 includes a comprehensive discussion of modality as presented in earlier literature. Van linden discusses different analyses of the modal domain, and summarises them in a useful table (p. 26). In addition to the basic modal categories (dynamic, deontic and epistemic modality), the chapter also includes a discussion of categories ‘on the edge of modality’: mood (indicative/subjunctive), volition, rational modality and evaluation. Van linden considers modality and evaluation to be partially overlapping categories, emphasising, however, that ‘not all evaluative categories express modal meaning’ and ‘not all modal expressions express evaluative meaning’ (p. 42).
Chapter 2 introduces the adjectives studied in the book. All the adjectives denote desirability, and they are divided into two classes according to the relative strength of desirability expressed by them: strong adjectives (e.g. critical, crucial, essential) and weak adjectives (e.g. appropriate, convenient, desirable). In addition to the difference in the degree of desirability, the categorisation is motivated in other ways as well. For example, strong adjectives are semantically bounded and combine with totality modifiers (e.g. absolutely essential), whereas weak adjectives are semantically unbounded and combine with scalar modifiers (e.g. very appropriate). Furthermore, weak adjectives can be used with if-clauses and gerundial complements, while strong adjectives are only found with that- and to-clauses. In this book, Van linden focuses on extraposition constructions, and so only that- and to-clauses are studied in detail.
Importantly, the two groups of adjectives also have different meaning potentials. In general, strong adjectives can only express dynamic and deontic meaning, as in (1) and (2). By contrast, weak adjectives can express both deontic meaning and non-modal evaluation, as in (3), but they cannot express dynamic meaning. Furthermore, strong adjectives generally take mandative complements only (i.e. ‘complements containing a potential SoA’; see (2)), while weak adjectives also take propositional complements (i.e. ‘complements containing a presupposed SoA’; see (3)). The difference between the two modal categories and the non-modal one can also be analysed in terms of factuality and the presence or absence of an attitudinal source: deontic and dynamic modality involve non-factual SoAs, whereas non-modal evaluation involves factual SoAs. On the other hand, deontic expressions and non-modal evaluations have an attitudinal source, while dynamic expressions do not have one. When it comes to the adjectives studied, weak adjectives always involve an attitudinal source, and depending on the factuality status of the SoA in the complement clause, they express either deontic or non-modal evaluative meanings. Strong adjectives, by contrast, express non-factual SoAs (with few exceptions). Depending on the presence or absence of an attitudinal source, they can express either deontic meaning or dynamic meaning, respectively.
Chapter 3 introduces the data and methods used in the diachronic analysis of modal adjectives. The adjectives studied have been chosen from several thesauruses, and their meanings fall within the positive sphere of the deontic modal domain. The diachronic data cover the entire history of English from Old English to Present-Day English. In addition to dictionary data, Van linden uses the following corpora to track the development of modal adjectival constructions: the York–Toronto–Helsinki Parsed Corpus of Old English Prose (750–1150), the Penn–Helsinki Parsed Corpus of Middle English, Second Edition (1150–1500), the Penn–Helsinki Parsed Corpus of Early Modern English (1500–1710), the Corpus of Late Modern English Texts (Extended version; 1710–1920) and the Collins COBUILD Corpus (c. 1990–5). In this chapter Van linden also provides frequency data for all the adjectives studied (p. 79). Later I will have some critical comments about the data used in the study, but for now I would just like to point out that some of the adjectives that Van linden discusses in detail in the chapters that follow are extremely infrequent in certain periods.
Chapter 4 focuses on the diachronic development of four strong adjectives that entered the English language in the Middle or Early Modern period: essential, vital, crucial and critical. Van linden shows that all four adjectives started out as non-modal items. For example, essential used to mean ‘being such by its true nature’, while crucial used to mean ‘cross-shaped’. The first semantic change of these adjectives involved the acquisition of relational meaning (e.g. heat is an essential property of fire). Later, they developed potential meaning, which is the prerequisite for dynamic modality (e.g. government is essential to formed and regular societies). Finally, through a process of subjectification (Traugott Reference Traugott1989), the adjectives came to denote desirability, and, consequently, they could be used in deontic expressions (e.g. it is essential that you come as soon as you can). In other words, Van linden argues that non-modal meanings are historically prior to dynamic meanings, which in turn are prior to deontic meanings. Although Van linden's argumentation in this chapter is otherwise solid and convincing, it must be pointed out that her data do not always provide compelling evidence to support her claims. For example, contrary to her hypothesis, the potential meaning of vital (‘indispensable for’) predates the non-potential meaning (‘constituting the essence of’) by two decades in her data. Similarly, the first attestation of the non-potential meaning of essential (‘constituting the true nature of’) is only two decades prior to the potential meaning. This could point to a simultaneous development of both meanings instead of two clearly separable stages. In any case, as Van linden herself also admits, more data from larger data sets are needed to add credence to her hypothesis.
Chapter 5 presents a diachronic study of the clausal complement patterns used with modal adjectives. Specifically, the focus is on the copular extraposition construction and transitive verb constructions (both active and passive), illustrated in (4), (5) and (6). The copular construction has been the more frequent one of the two from OE to PDE, but there was a peak in the frequency of the transitive construction in EModE.
(4) It is important to be realistic about your prospects. (copular)
(5) We think it important that the Pope make it a public invitation without any prior notice to the North Vietnamese. (active transitive)
(6) It was considered essential that the students complete the test. (passive transitive)
As can be seen from examples (4)–(6), the clausal complements of the modal adjectives are either to- or that-clauses. Van linden shows that while that-clauses were more common in Old English, to-clauses started to gain ground in Middle English, and in Present-Day English to-clauses constitute two-thirds of all mandative clausal complements. Van linden argues that the rise of the to-infinitive in the Middle English period can be explained by analogy with similar changes in mandative constructions with intention and manipulative verbs (Los Reference Los2005). However, while the frequency of that-complement clauses with verbal matrices continued to decline up to 1800 (Rohdenburg Reference Rohdenburg1995), the proportion of that-clauses in adjectival constructions has remained relatively stable since the Late Modern period. Van linden proposes that this can in part be explained by lexical determination (e.g. both vital and crucial favour that-clause complements), while stylistic factors (horror aequi) and information structure (the informational salience of the subject of the that-clause) may also be relevant. There is also a distributional difference between mandative and propositional complements: while mandative complements are more likely to be expressed by to-clauses in PDE (3:1), the distribution of that- and to-clauses is equal with propositional complements.
Chapter 6 focuses on the development of non-modal evaluative constructions from deontic constructions. Before proceeding to a more detailed analysis of the diachronic data, Van linden discusses the semantic properties of complement relations in functional terms (see Cristofaro Reference Cristofaro2003: 109–10). She distinguishes between three types of complement-taking predicates (CTPs): modal, desiderative and commentative, which correspond to dynamic modality, deontic modality and non-modal evaluation, respectively. Van linden argues that commentative predicates pertain to a different level of clause structure from modal and desiderative predicates, and they can also be set apart from modal and desiderative CTPs in terms of time-reference, discourse dependency and semantic integration. More specifically, commentative predicates are time-reference independent and discourse dependent, and they involve no semantic integration. By contrast, the only difference between modal and desiderative CTPs involves the degree of semantic integration, which is high with modal CTPs and low with desiderative CTPs. Van linden suggests that the similarity between modal and desiderative CTPs can explain the similarities of dynamic and deontic expressions.
The case study in chapter 6 involves two sets of adjectives: ‘importance adjectives’ (important, essential, crucial) and ‘appropriateness adjectives’ (appropriate, proper, fitting), which can be used to express both deontic and non-modal evaluative meaning in Present-Day English. Van linden shows that the non-modal uses of these adjectives developed out of deontic uses. However, the path of development was different for the two groups studied. Adjectives of importance developed non-modal evaluative meanings through secondary complementation, where the primary (deontic) mandative complement was followed by a secondary (non-modal) propositional complement. The mandative complement was a verb of cognition or verbalization, such as remember, observe or note, which guided the reader's/hearer's mental focus on the secondary proposition, as in (7).
(7) It's also important to remember that children change as they grow up.
Van linden suggests that before developing a purely evaluative use, adjectives of importance were used in a pattern where the mandative complement was not overtly expressed but still pragmatically implied.
(8) I'm not into just designing for those people with money. I mean I think it's really important that I want to reach as broad a field as possible.
The construction in (8) bears close resemblance to the mental focus construction in (7). Importantly, the speaker does not provide an evaluation of the proposition in the that-clause; rather, she wants to focus the hearer's attention on the importance of the proposition. In PDE, adjectives of importance can also be used in true evaluative expressions, as in (9).
(9) I think it's so terrific that you rang, it's so important that you rang because your experience is far more common than people will be prepared to admit to you.
The that-clause in (9) expresses a presupposed propositional complement, which the speaker assesses to be terrific and important. Therefore, (9) is an example of genuine non-modal evaluation, and it cannot be read as a mental focus construction.
Similarly to the adjectives of importance, the appropriateness adjectives were also first used with mandative complements. Contrary to the importance adjectives, however, they developed evaluative uses through bridging contexts, where the factuality status of the SoA expressed in the complement clause was ambiguous between potential and actualized readings.
Van linden draws the diachronic studies together by pointing out some general trends. First, modal adjectives started their development from non-modal lexical items and later developed dynamic, deontic and non-modal evaluative meanings (in this order). She also points out an interesting implication: if a strong adjective can express deontic meaning, it can also express dynamic meaning. Likewise, if a weak adjective can be used to express non-modal evaluation, it can also be used to express deontic meaning.
In chapter 7 Van linden provides a brief introduction of the data and methods used in chapter 8. The data comprise a total of 12 weak adjectives and 10 strong adjectives, which Van linden analyses by using a multiple distinctive collexeme analysis (Stefanowitsch & Gries Reference Stefanowitsch and Gries2003). To control for false positives, she applies the Bonferroni correction to the results.
Chapter 8 presents a detailed description of modal and non-modal constructions in PDE from a constructional perspective. Van linden proposes that certain evaluative patterns, such as the locative and knowledge/acquisition of knowledge patterns, can be considered to be partially filled constructions, where ‘some aspect of the form . . . or the meaning/use are not strictly predictable from the component parts or from other constructions already established to exist in the language’ (Goldberg Reference Goldberg, Brown and Miller1996: 68). She also provides a more detailed discussion of the problems related to the ambiguities in deontic and dynamic readings, arguing that the potential ambiguities arise from the fact that ‘it is [impossible to demonstrate] the absence of an attitudinal source in extraposition constructions’ (p. 310). In other words, as it is almost always possible to argue for a presence of an attitudinal source, a deontic reading can never be ruled out with certainty.
Chapter 9 provides a conclusion to the book. Here, Van linden recapitulates the most important findings of her work and offers some final points of clarification. She emphasises that the conceptual map that she has drawn on the basis of her data is only intended to apply to English: ‘it is not assumed to have universal relevance’ (p. 321). Nevertheless, she suggests that it would be interesting to see whether the arguments presented in this book could also have cross-linguistic validity.
Van linden's book is impressive for many reasons, and it is not possible to present all her insights in this review. She not only provides a convincing account of the meaning and use of modal adjectives in PDE but also studies their diachronic development all the way from OE to PDE. The pathways of change suggested for the different sets of adjectives (weak vs strong, importance vs appropriateness) are plausible, and they are explained by general mechanisms of semantic change, such as metaphorisation and subjectification. Moreover, by paying close attention to the complement types in the extraposition constructions, Van linden connects the changes in adjectival complementation to the more general changes in clausal complementation, which lends even more credence to her claims.
My only concern relates to the data used in the case studies. As I already pointed out, the token frequencies of some of the adjectives studied reveal that the hypothesis concerning the general pathway of semantic change is at times based on very limited data. For instance, the earliest attestations of critical and crucial come from the Early Modern period, but Van linden's Early Modern data only include six tokens of each adjective. This raises two questions: is the hypothesis concerning the path of semantic change from pre-modal meanings to dynamic and deontic meanings sufficiently supported by the data? Second, are there enough data to allow for an accurate dating of these changes?
To illustrate these problems, let us consider essential and fitting, which were both discussed in the case studies. The earliest deontic use of essential in Van linden's data dates back to 1842. However, I would suggest that a deontic reading is possible, and even probable, in both (10) and (11), which have been taken from the Eighteenth Century Collections Online (ECCO).
(10) It is not the Business of this Tract, to enquire, what are the especial Grievances of this Article, and why a Nation, to whose Interest it is so essential to prevent the Exportation of their Wooll, should at the same time have the Laws for that purpose so ill Executed, as that in almost every part of the Kingdom the Wooll is carried Abroad, not with allowance, we cannot say that, but with an open Violence . . . (ECCO, 1713)
(11) I am still afraid that Fischer set out too late: It is, however, very important, and very essential that we should raise large contributions. I see no other resource for our most urgent expences . . . (ECCO, 1765)
Similarly, Van linden found the earliest unambiguous examples of non-modal evaluations with fitting only after 1850. However, I would argue that example (12) from 1646 is a genuine instance of non-modal evaluation (taken from the Early English Books Online).
(12) It is not an uncertainty that places must be as farre distant in scituation, as in use: Therefore Abraham saith, That there is μέγα χάσμα, a great gulfe, or gap between Lazarus and Dives. So David distinguished between the height of heaven, and the deepe of hell; so doth Amos, and Esay: and it's fitting that heaven and hell, the saints and the wicked, the joys of the one, and torments of the other, be as remote as may be . . . (EEBO, 1646)
In (12), the author argues that the Earth is the centre of the universe and that it is fitting that the saints in heaven and the wicked in hell are as far removed from each other as possible. The complement that-clause refers to an existing situation, and it is therefore propositional, not mandative. If this reading is accepted, then (12) represents an evaluative use of fitting that predates the first attestation in Van linden's data by over two hundred years.
Examples (10)–(12) show that at least some of the developments described by Van linden took place (or started) much earlier than her data suggest. It remains to be seen whether such examples also affect Van linden's general line of argumentation. I must admit that I would be a little surprised if they did, but it must nevertheless be acknowledged that the examples from the ECCO and the EEBO illustrate the need for further research with larger data sets.
To conclude, Van linden's book is a significant addition to the research on modality and clausal complementation. Its numerous merits far outweigh its few weaknesses, and I especially recommend it to anyone interested in modality, semantic change and clausal complementation.
