The World Englishes (WE) paradigm stresses the global heterogeneity of English as it distinguishes between different Englishes, including those varieties spoken as a native language (ENL), as a second language (ESL) and as a foreign language (EFL). In the discussion of his foundational model of WE, Kachru (Reference Kachru, Quirk and Widdowson1985) strongly focuses on ESL varieties and acknowledges differences between Englishes as innovations. While these innovations in ESLs have become the focus of much research in the field, many studies follow a deficiency approach: ESLs are defined negatively in contrast to British and American English as norm-providing yardsticks.

In Corpus linguistics and sociolinguistics: A study of variation and change in the modal systems of World Englishes, Beke Hansen moves away from this approach and shifts to a positive definition of ESLs, which focuses on variety-internal variation and substrate influences. She analyzes modals (must) and semi-modals (have to, have got to and need to) of obligation and necessity across and within two ENLs, American and British English, and three ESLs, Hong Kong, Indian and Singaporean English. For this study, she takes a corpus-based sociolinguistic approach that uses the metadata of the three components of the International Corpus of English (ICE) to highlight cross-variety differences and variety-internal variation with regard to language-internal and -external factors. Much of her research focuses on diachronic variation via an apparent-time approach as she investigates whether ESLs follow the general trend in the decline of core modals and the rise of semi-modals, which has been described for ENLs.

The book has an impressive size: 345 pages of text plus a 66-page appendix. It is organized in eleven chapters: the core empirical part consists of five chapters, which are preceded by three theoretical chapters and a chapter on data and methods. The book closes with two concluding chapters which discuss the results and the methodological implications in the context of the wider field of WE.

Hansen's theoretical part is overall very detailed and opens with a definition of modality and modal verbs in the introduction. Two binary distinctions are at the core of her analysis: deontic (or root) vs epistemic modality; and modals (must) vs semi-modals (have to, have got to and need to). She then defines the main sociolinguistic variable of her study, deontic modality, and the four variants, must, have to, have got to and need to. These definitions and distinctions are illustrated with many examples from her corpus data.

Chapter 2, ‘Previous research’, discusses a wide range of studies on modals and semi-modals of obligation and necessity across WE and concludes with a description of a postcolonial research agenda. Hansen first gives an overview of corpus-linguistic work on modals of obligation and necessity in ENL varieties with a focus on diachronic changes: she highlights a general decrease in core modals (must) and an increase in the semi-modals (have to, need to) as well as regional variation, particularly with regard to have got to. The literature review on modal systems in ESL varieties is broader, as many studies on different modals are described in great detail. Hansen takes a very critical approach to discussing previous research: she criticizes an over-reliance on listing and comparing frequencies of forms without taking into account their functions. Hansen also calls for a widening of corpus-linguistic analyses to include language-internal as well as language-external factors, such as text type, age, gender and L1 of the speakers. She thus advocates a sociolinguistic approach to describing variation in corpus data in closer detail. She is especially critical of interpreting synchronic frequencies in terms of diachronic trends, as there is an underlying (unproven) assumption that the current states of British and American English are the ultimate goal of development. The last section focuses on this Western-bias in corpus-linguistic research on ESLs. Hansen rejects a negative definition of ESLs in contrast to ENLs and advocates a positive definition of ESL, which focuses on variety-internal variation and substrate influences.

In contrast to this critical discussion of previous research, the discussion of ‘The theoretical framework’ in chapter 3 is purely descriptive. Hansen first presents Kachru's (Reference Kachru, Quirk and Widdowson1985) circles model and Schneider's (Reference Schneider2007) Dynamic Model of Postcolonial English. She then utilizes these models to discuss the socio-historical background of English in Hong Kong, India and Singapore, highlighting salient differences between these three ESLs. Hansen then introduces Mufwene's (Reference Mufwene2001) Feature Pool model, including different selection processes – as introduced by Biewer (Reference Biewer2015) – which govern speakers’ choices from the pool, as a toolkit to analyze the structure of these three ESL varieties. Hansen thus unites a wide range of different theories to discuss the variation in the modal systems within and across different Englishes.

Chapter 4, ‘Methodology’, presents Hansen's approach, which she describes as sociolinguistic corpus-based. She uses the ICE corpora but moves beyond counting and comparing frequencies of forms in different (sub)corpora toward a variationist analysis of the realization of modality in and across different Englishes. The variationist study uses the ICE metadata and includes a wide range of language-internal and -external predictor variables. This corpus-based analysis is then supplemented with a questionnaire study, which investigates the perception of speakers toward variation in the modal system. However, this extremely broad approach is not applied to all varieties under investigation. Instead, the study zooms in more and more on specific varieties. In a first traditional corpus-linguistic step, frequencies of forms and their function are compared across the five varieties; the main part of the variationist approach is applied to British, Hong Kong and Indian English, and the questionnaire focuses on the perceptions of 53 Indian informants. The analysis of variation in the modal systems with regard to the Feature Pool model is not explained in the methods chapter. In contrast, the ICE data used for the study are discussed in great detail: in particular, Hansen highlights the sociolinguistic make-up of the corpora under investigation and comparability issues in ICE.

The empirical part consists of five chapters: three shorter chapters focus on the corpus-derived results on the use of must, have to, have got to and need to and their constraints across and within different Englishes. Two more theoretical chapters discuss the Feature Pool model and selection processes of the realization of modality of obligation and necessity in the three ESLs.

Chapter 5, ‘Obligation and necessity in ENL and ESL’, provides an overview of regional variation of modality on the highest level of abstraction by presenting the overall frequency distribution of must, have to, have got to and need to and their functions (deontic vs epistemic modality) in British, American, Hong Kong, Indian and Singaporean English. On the one hand, the analysis shows strong similarities across these five Englishes: must and have to are the main competitors, while have got to and need to are rather marginal. Furthermore, speakers across all varieties predominantly use must to express epistemic modality, whereas semi-modals are rarely used in this sense. On the other hand, there are salient cross-variety differences: the three ESLs are marked by a saliently higher overall frequency of all four modals and semi-modals. Beyond these overall trends, Hansen concludes that all varieties have a distinct usage profile, which ‘results in a complex web of relationships between the varieties’ (p. 158).

The following two chapters investigate the different input sources and selection processes along the Feature Pool model that have led to the state of modality in the three ESLs illustrated in the previous chapter. Chapter 6, ‘The Feature Pool of obligation and necessity’, describes the different input varieties and languages that have contributed to the Feature Pool of Hong Kong, Indian and Singaporean English. Hansen analyzes (1) the state of modality in different periods of British English as the historical input variety, (2) modality in a wide range of different substrate languages from the three countries and (3) American English as a globally dominant variety, which theoretically also affects the three ESLs. All three aspects of the analysis are on a high level of abstraction. Definite links between these theoretical considerations and regional distinctions are difficult to establish but Hansen is still able to formulate certain hypotheses about the effects of the different sources. Chapter 7, ‘Selection principles in the feature pool’, then discusses four selection principles: (1) substratal transfer takes into account the multilingual settings of ESLs; (2) cultural motivations shed light on the local appropriation of English; (3) Second Language Acquisition processes are a major distinguishing factor between ENLs and ESLs; and (4) cognitive principles of speech production and processing are described as having a universal effect on all varieties. With this approach Hansen not only tries to explain the processes which have led to the variability in the modal systems of obligation and necessity but also current constraints of variation. For example, she shows that the frequency of epistemic must is higher in varieties of English that have developed in contact situations with languages that are typologically similar to English in that they also encode epistemic and deontic modalities in one form. This approach is a very broad one and tries to highlight the web of interactions between and within different Englishes, which is the ultimate conclusion of the preceding chapter. However, this approach often remains largely theoretical as it is difficult to operationalize these different factors separately and to pinpoint individual effects unambiguously.

Chapter 8, ‘Apparent-time development’, presents a diachronic perspective on modals of necessity and obligation in ICE Hong Kong and India using an apparent-time study. This chapter focuses on internal variation in the two ESLs and at the same time checks in how far changes in the modal system in these two ESLs align with changes in the use of modals reported for ENLs. The apparent-time results show that the modal systems of both ESLs are undergoing language change and follow the decline of core modals and the rise of semi-modals: must is decreasing from the oldest to the youngest age group, while have to is on the rise with a peak among the youngest age group. This rise of innovative have to follows the well-known gender pattern as females lead this rise from below. No clear changes in frequencies are detectable for the semi-modals have got to and need to. The analysis mainly uses descriptive statistics, which are very well done as Hansen normalizes the raw token frequencies by the number of words spoken by the speakers in the different age groups. This chapter is concise and highlights the potential of the ICE metadata, which unfortunately is not available for ICE Singapore. With her apparent-time study Hansen approximates a much-needed diachronic perspective on ESLs. However, real-time evidence remains a desideratum.

In order to further entangle the web of interactions that constrain the modal systems of obligation and necessity, Hansen moves to a multivariate analysis of must vs have to using binary logistic regression modelling in chapter 9, ‘Competition between must and have to’. She first investigates the effect of a range of language-internal and -external factors on must vs have to in the spoken components of ICE Great Britain (GB), Hong Kong and India. This local perspective is enhanced by a global one by means of one regression model that draws on the complete data set with all tokens of must and have to in their deontic function from all five varieties (written and spoken). This combined approach shows similarities and differences across the different Englishes. The global perspective highlights the fact that Hong Kong English is the only variety that is significantly different from British English due to a particularly high frequency of have to. This approach then shows the mainly uniform effects of verb semantics, type of subject and text type. Particularly important for Hansen's argument is the correlation between the formality of a text type and the use of must: formal text types (e.g. academic writing) favor must and informal ones (e.g. private dialogues) favor have to. The part of the analysis which focuses on language-external variation highlights salient cross-variety differences: age has a significant effect on the distribution of must vs have to in ICE Hong Kong and India, where must is favored by older speakers, but not in ICE GB. Gender only has a significant effect for ICE India: there is a preference for must among males. With regard to text type, the probability of must in public dialogues is significantly higher in the two ESLs than in ICE GB. Hansen explains this finding via cross-cultural differences: the direct expression of obligation in public speeches is marked in societies with an independent construal of self (i.e. Western societies such as Britain), but it is unmarked in societies with an interdependent construal of self (i.e. Hong Kong and India). Further cultural differences emerge from the varying effect of type of subject: in ICE India, must correlates with second-person non-generic uses. Hansen explains that this imposition of a direct obligation on a hearer by the speaker is rare in Western cultures as it runs counter to the principle of democratization of discourse. This is confirmed by the ICE GB data: have to is preferred for second-person non-generic uses and must correlates with first-person non-generic subjects, i.e. self-imposed obligation by the speaker.

The chapter then closes with one of the book's major conclusions, based mainly on evidence from chapters 8 and 9. These variationist analyses have shown that the modal systems of both ESLs are changing: the core modal must is declining and the semi-modal have to is on the rise. Age has emerged as a significant factor in ICE Hong Kong and India but not in ICE GB. Hansen thus concludes that language change toward have to is near completion in ICE GB but still well under way in the two ESL varieties. Based on slightly different age and gender patterns in the variation, Hansen further distinguishes between ICE India, where language change is in full swing, and ICE Hong Kong, where it is nearing completion. Thus all varieties undergo the same restructuring process from must to have to but at different speeds. In this line of argumentation, this change does not spread from a ‘center’ (British or American English) to the ‘periphery’ (ESLs) but is a universal one. For Hansen this change has its origin in historical variation in British English and starts in informal spoken language and then proceeds to increasingly more formal text types, as shown by the correlation of formality of text type with the distribution of must and have to. Thus, regional differences, including the complex web of relationships between the varieties, are sidelined by universal notions of language change.

Chapter 11, ‘Thematic conclusion of the study’, continues with the discussion of variation and change in the modal system(s) of WE. Hansen discusses variation in the modal systems mainly with regard to different substratal influences but stresses that there is only one modal system: all varieties undergo the same restructuring process, they are just at different stages of this process and are influenced by different substrates. She concludes that WE research needs to move away from highlighting differences and instead should focus more on similarities. This approach could be beneficial to understanding language change in a wider perspective.

The last chapter of the book, ‘Methodological implication of the study’, focuses on the pros and cons of different types of corpora for linguistic research. Hansen first highlights the benefits and limitations of small vs large corpora with regard to corpus size, diversity of design, representativeness and tidiness. As text type is essential for understanding the variation in the modal system, she concludes that a diverse design outweighs size for her study. She then highlights the potentials and benefits of rich data, i.e. the data contain information on the speakers and the language the data represent is not dissociated from the user. The productive use of the richness of the ICE data is a unique selling point of Hansen's work. She uses the metadata, such as age and gender, very effectively and never loses touch with speakers in all of her complex analyses as she highlights cultural differences, substratal effects and selection processes that govern the speakers’ choices in close detail. Such an approach, which applies sociolinguistic methods to available corpus data, offers unexplored possibilities for the field of WE but is in need of rich data.

Hansen presents an impressive work on variation within and across Englishes worldwide. She unites different theories and approaches to paint a rich and detailed picture of language variation and change in the modal systems of WE. Despite the richness and diversity of her analysis she does not treat language as detached from the speakers: for example, in a total of 187 excerpts from her data, she presents the voices of the speakers who make the corpora. While the sheer length and detail of the book can be a bit overwhelming, Hansen includes summaries at the end of every chapter to make sure that the reader is not lost in all the detail. Hansen's work is very important and timely for the field of WE, as she shows how to use small and rich corpora to their full potential in an age of ever bigger corpora.