This book is a thematic collection or articles addressing the interface of English as a Second Language (ESL) and English as a Foreign Language (EFL). As such, it follows a string of publications that have recently taken on the task of bridging the ‘paradigm gap’ (Sridhar & Sridhar Reference Sridhar and Sridhar1986) between indigenized varieties and learner English research from diverse perspectives. The endeavor continues to be timely and relevant, particularly when viewed in terms of linguistic creativity, which in English linguistics tends to be seen as the privilege of specific groups of language users: English as a Native Language speakers (ENL), conventionalized New English speakers (ESL) and English as a Lingua Franca speakers (ELF). As for EFL speakers or learners, any deviations from the native-speaker (standard British or American) norms are traditionally interpreted as lack of proficiency or a failure to reach the set target, not creativity. As the editors point out in their introduction, however, many recent studies indicate that learners and EFL speakers, too, are increasingly credited with strategic and functional creativity as are ESL or ELF speakers, be it in the domains of code-switching, lexicon, phraseology or morphosyntax. The World Englishes paradigm shift is gradually proceeding to the expanding circle. Hence, rethinking linguistic creativity, as the title suggests, continues to be called for. The corpus linguistic methodology utilized throughout the book is particularly valuable for drafting a quantitative framework for what is meant by linguistic creativity, thus far often examined in qualitative studies. The volume consists of the editors’ introduction and eight research articles. It is a republication of the 2016 special issue of the International Journal of Learner Corpus Research 2(2).

A central topic of discussion in the articles concerns the concepts of error and innovation and the definitions of the two. While articles examining ESL and EFL varieties form the majority in the book, I am pleased that it also includes a contribution on linguistic innovations in ELF (Brunner, Diemer & Schmidt), because it turns out that defining an innovation is a very different issue in the two types of contexts. Speakers’ ability and relative freedom to utilize their linguistic resources in flexible, non-normative ways in ELF interaction is one of the central tenets of the paradigm; instead of errors, ELF scholars see pragmatic and multilingual competence (e.g. Jenkins Reference Jenkins2006; Seidlhofer & Widdowson Reference Seidlhofer, Widdowson, Albl-Mikasa, Braun and Kalina2009). It is therefore not surprising that Brunner et al.’s view of innovations is broader and based on their functional appropriateness and general acceptability in conversation. Whether or not they stabilize in use is of little relevance, as ELF communities tend to be highly transient. The studies focusing on linguistic creativity in ESL and EFL contexts, however, are centrally concerned with distinguishing between errors, generally seen as ephemeral learner language features, and innovations, i.e. features which go through propagation and become conventionalized in the speech community. There is by now a long tradition of identifying the linguistic processes leading to the emergence of new, established features in institutionalized varieties (e.g. Williams Reference Williams1987; Schneider Reference Schneider2012). Thus carrying out usage-based corpus studies on how – and what kinds of – individual-level errors or creative expressions lead to conventionalized innovations is of obvious relevance as well.

Something that I was left pondering about in the book's discussions on error versus innovation is a certain lack of problematization concerning the norm. Errors and innovations can only be identified as such in comparison to some form of normative and established language use. This is defined by Callies (p. 100) as ‘the main standard varieties (British and American)’, and the benchmark corpora used by many other authors indicate a similar stance. Errors, then, are described, for example, as ‘unacceptable by native speakers’, ‘deviant use of the norm prevailing in a given speech community’ (Deshors, Götz & Laporte, pp. 3, 11) or ‘low on acceptability in native varieties’ (Horch, p. 153). Of course, the book is primarily concerned with the middle ground, i.e. endonormativity: the ‘given speech communities’ setting their own norms through conventionalized linguistic innovations. The above types of definition nevertheless lean quite heavily on the (British or American) native-speaker norm. In the meantime, English language teaching in many EFL countries is slowly coming to terms with the plurality and global nature of English and starting to gear towards English as an international language (e.g. Alsagoff et al. Reference Alsagoff, Sandra Lee, Hu and Renandya2012) and ELF awareness (e.g. Sifakis & Bayyurt Reference Sifakis, Bayyurt, Jenkins, Baker and Dewey2018). Communicative competence and plurilingualism have already become major foci in language curricula, partly overriding grammatical accuracy as the first and foremost goal of acquisition. I realize that pedagogical considerations fall outside the focus of the book, but as an English linguist responsible for educating future English teachers in an EFL context, I could not help but wonder if the concept of learner English itself – also in variationist linguistics – might be in need of some rethinking. For example, in the article by Edwards & Lange (p. 142), some of the usage patterns discovered in the data are described as illustrative of ‘an interlanguage strategy, whereby learners latch onto particular phraseological crutches to compensate for the absence of more varied linguistic resources’ (my italics). An ELF researcher would formulate this thought in very different terms, as would many present-day SLA researchers (see, e.g., Douglas Fir Group 2016).

The above comment is by no means intended as criticism of the studies presented in the book, but rather as a gentle observation on the covert persistence of the deficiency mindset when discussing learner language. The articles themselves are a definite step up on the subjects of innovation and linguistic creativity in global Englishes. To begin with, editors Sandra Deshors, Sandra Götz and Samantha Laporte give a comprehensive introduction to relevant topics and state-of-the-art research on linguistic creativity across the ESL and ELF paradigms, debating the fine line between errors and innovations in light of English variationist research. They appear to have made a deliberate decision not to venture into the field of SLA research to avoid a potential quagmire of definitions and perspectives. Although a justified choice, this has the downside of presenting SLA research in rather superficial terms at times. As Deshors et al. point out, however, integrating EFL and learner English into the linguistic creativity discussion is currently making considerable headway through empirical, methodological and theoretical advances, all of which are illustrated in the present volume. Focal to the emergence of innovations is the diachronic and community-based trajectory examined in section 4. This aspect clearly distinguishes the ESL/EFL studies in the book from ELF research, which is presented as a parallel and hence relevant field of study.

In ‘This hair-style called as “duck tail”: The “intrusive as”-construction in South Asian varieties of English and Learner Englishes’, Christopher Koch, Claudia Lange & Sven Leuckert present a model example of a comparative ESL–EFL corpus study by examining ‘intrusive as’ (e.g. to call X as a fool) in South Asian Englishes versus a range of learner Englishes. The corpora comprise the South Asian Varieties of English newspaper corpus (SAVE) and two learner corpora, the International Corpus of Learner English (ICLE) and the International Corpus Network of Asian Learners of English (ICNALE), where the informants are university students and upper-intermediate to advanced learners of English. SAVE was annotated and processed to find appropriate complex-transitive instances and the findings were compared against British and American native-speaker data to create a list of sixty verbs with ‘intrusive as’ in Asian Englishes. The same kind of methodological detail, typical of variationist corpus linguistics, can be witnessed across all articles of the book. Six frequent verb lemmas are studied in the corpora against six context variables, indicating that term, rather than the stereotypical call, presents fairly established use of ‘intrusive as’ in South Asian Englishes. Central variables favoring the choice are, for example, active voice and a long-distance/heavy NP between the main verb and the complement. The learner corpora display lower relative use of ‘intrusive as’ with the exception of some Asian and Turkish learners. The conditional inference tree method used in the analysis nevertheless does not rate language background (apart from Turkish) very high as a predicting factor. Rather, the distance from the main verb is again found to be significant. The authors conclude that ease of processing emerges as a motivating factor in both ESL and EFL data sets, thus finding similarities rather than differences across the variety types.

In the next article, ‘Detecting innovations in a parsed corpus of learner English’, Gerold Schneider & Gaëtanelle Gilquin delve into the lexico-grammatical interface often credited as a particularly productive area for innovations in New Englishes (e.g. Schneider Reference Schneider2004). The authors focus specifically on prepositional overuse in verb/adjective + PP combinations to try to examine the error vs innovation borderline in the syntactically parsed International Corpus of English (ICE) subcorpora and ICLE, using the British National Corpus (BNC) as a native-speaker point of comparison. Following Van Rooy's (Reference Van Rooy, Mukherjee and Hundt2011) criteria of systematicity and acceptability, innovations are operationalized as collocation measures and frequency of co-occurrence across different learners and L1 backgrounds. Frequency-based collocation statistics, O/E and T-score, are used as the tools of analysis; one of the aims of the article, in fact, is to test their applicability in this research context. The two methods turn out to produce different but complementary results. A comparison of the ESL and EFL corpora shows, for example, that the former have a tendency for analogy through noun complementation (e.g. discuss[ion] about, [to place] stress on), while the latter utilize to as a generic preposition to some extent (e.g. assist to, impose to). Examining similar occurrence patterns across the variety types, on the other hand, leads the authors to conclude that the present methods do not allow identification between errors and innovative usage patterns.

This feat is, nevertheless, accomplished by Bertus Van Rooy & Haidee Kruger in their study on the progressive form in Black South-African English (BSAfE): ‘The innovative progressive aspect of Black South African English: The role of language proficiency and normative processes’. The writers have the advantage of a set of learner data, from grade 10–11 school essays to dissertation extracts, which presents a continuum of proficiency and allows the identification of learner language features in the low-proficiency data versus more stable innovations in the advanced ones. As a variety, BSAfE straddles the ESL–EFL divide and presents the perfect opportunity to witness conventionalization in action. Van Rooy & Kruger use samples of unedited and edited published academic writing by BSAfE authors, i.e. local endonormative texts, as their points of comparison. Of the features under investigation, omission of auxiliary be is found to fade out towards the high-proficiency texts and it therefore presents the clearest case of a learner language feature. Various semantic uses of the progressive form are categorized into four subgroups according to their acceptability in native-speaker grammar descriptions. Of the non-standard usages, the perfective is particularly characteristic of the lower-proficiency texts, while usages which are maintained also in academic writing are of the borderline standard type expressing (non-delimited) ongoingness. As (delimited) ongoingness is one of the central aspectual properties of the standard progressive form as well, it is not surprising that it becomes extended and conventionalized in a contact variety such as BSAfE. Van Rooy & Kruger also find that this feature is not subject to editorial intervention, unlike perfective usages.

Marcus Callies focuses on the processes underlying lexical innovations, specifically features of derivational morphology, in ESL and EFL corpora, in his article ‘Towards a process-oriented approach to comparing EFL and ESL varieties: A corpus-study of lexical innovations’. The data arise from components of ICE (student writing), Global Web-based English Corpus (GloWbE), ICLE and ICNALE. Potential comparability issues and their impact on the results are discussed at length. Corpus searches include both Germanic and Non-Germanic prefixes and affixes, as the stem allomorphy and morphophonological changes caused by the latter tend to present learners with additional problems. The potential non-normative instances are then checked against native-speaker data. The broad categories emerging from the data are those of cross-linguistic influence – mainly observed in EFL data from L1 Spanish and Italian learners – word coinage using existing English resources, and cognitive strategies. (Over-)regularization is more of an EFL strategy, while over-affixation is observed more in the ESL data, but overall, cognitive processes underlie both variety types. There is also considerable variation within the categories. Callies predicts that the novel forms are not likely to gain acceptance in exonormative EFL contexts, but of course, there is a difference between learner and user settings. Acceptance in ELF use is quite common (e.g. Reference SchneiderPaulasto forthcoming), and Callies suggests that the Internet may yet provide an opportunity for some degree of conventionalization as well.

Alison Edwards & Rutger-Jan Lange direct their attention to frequently used academic English phraseology in Inner, Outer and Expanding Circle Englishes in ‘In case of innovation: Academic phraseology in the Three Circles’. Their data arise from the academic writing sections in eight ICE corpora and the Corpus of Dutch English (NL-CE) compiled by Edwards, one of the few EFL corpora comparable to ICE at present. The corpora were searched for all three-word clusters (3-grams), which were limited by frequency to 197. These are normalized per author to observe individual variation, using ANOVA, F-tests and chi-square tests for the statistical analysis. The authors find that East African corpora stand out with high total and normalized frequencies, while NL-CE presents no significant overuse or overreliance on specific cluster types, which might be indicative of learner language. A detailed study of four clusters (the fact that, on the other hand, due to the, in case of) reveals differences: the first two clusters appear to be particularly frequent in specific authors’ writing, indicative of a learner strategy, while the frequent use of due to the is common to all non-native-speaker corpora, including NL-CE. This implies a higher acceptability in these varieties. Edwards & Lange conclude that 3-grams are not ideal for assessing differences between ESL and EFL variety types. Rather, they illustrate the problems of the dichotomy: variation needs to be considered in more complex terms, including geographic proximity, proficiency, register variation and cognitive processes.

In ‘Innovative conversions in South-East Asian Englishes: Reassessing ESL status’, Stephanie Horch takes a different kind of approach to assessing innovations, focusing on verb-to-noun conversion (VNC) in two South Asian varieties, Singapore English (SgE) and Hong Kong English (HKE). The two varieties are considered to have reached different phases of the Dynamic Model and hence have different linguistic properties despite their partly similar historical backgrounds. The data arise from GloWbE, with the BrE component as a point of comparison. Horch selected a total of twenty verbs with potential for conversion, and the corpora were searched for samples of 1,000 instances, which were then coded manually and analyzed using logistic regression. The results show statistically significant differences between the Asian and British varieties and especially high frequencies for HKE. Horch interprets the HKE result as a learner language strategy, with frequently used nouns (e.g. examination) blocking VNC. Moreover, conversion emerges in the formal register in HKE, while in SgE it is confined to more informal blog posts. This shows that the high institutionalization of SgE leads to VNC fading out, i.e. there is little evidence of propagation. The feature thus compares to the omission of be in progressive constructions in Van Rooy & Kruger's study, and so, Horch makes a valid observation on the relevance of the Dynamic Model in assessing the level of conventionalization.

Anna Rosen focuses on a lesser-known variety, Jersey English (JersE), and its French-induced characteristics in comparison to French learner English (‘The fate of linguistic innovations: Jersey English and French learner English compared’). The study highlights the role of regional identity in the maintenance of JersE dialect features, which are in the process of levelling through mainland British influence. The spoken data arise from Rosen's sociolinguistic interviews and the LINDSEI French subcorpus. Three features are under investigation: verb-and-infinitive (e.g. I went and see…), existential there's with time reference (e.g. there's a few years that…) and the discourse particle eh. In the JersE data, all are mainly associated with bilingual speakers and monolingual English speakers with a strong local attachment. Rosen finds that of the three features, eh is most likely to be maintained as a regional identity marker, while the first two have a very low profile. Existential there's with time reference is not even used by the LINDSEI informants, while the other two emerge as learner features as well. Diverse social factors are integral to their use and survival in JersE, however.

The ELF scholars Marie-Louise Brunner, Stefan Diemer & Selina Schmidt utilize transcribed sections of the Corpus of Academic Spoken English (BabyCASE), comprising academic Skype conversations between speakers of different L1s, in their study of lexical innovations entitled ‘“It's always different when you look something from the inside”: Linguistic innovation in a corpus of ELF Skype conversations’. Compared to Callies, they take a more comprehensive approach and include, e.g., code-switching, semantic approximations and idiomatic expressions, utilizing manual methods in addition to corpus searches. After the exclusion of ‘non-innovative deviations’, the broad categories are L1 influence, approximations and ad hoc innovations, where approximations roughly correspond to what Callies considers cognitive strategies and ad hoc innovations concern the utilization of English word formation processes (also Reference Paulasto, Kolehmainen, Riionheimo and UusitupaPaulasto forthcoming). The findings are examined against multilingualism and ESL research to an extent. The article essentially illustrates the breadth of linguistic innovations in the data at the lexical level, providing a template for further research and adding an important theoretical and empirical perspective to the book.

In conclusion, the articles in this volume provide multifaceted and thought-provoking perspectives on linguistic creativity. The overarching theme is the formal similarity of many innovations in ESL and EFL/learner English versus their disparate conceptualization, and the need to look beyond such borders using novel quantitative and corpus-based methods. In this, the book definitely succeeds. It also becomes evident that in order to push the boundaries further, new and innovative corpora are in a key role. Public, easily accessible and annotated corpora such as ICE, ICLE and LINDSEI are certainly useful and provide mutually comparable data, but in order to look into the emergence and possible conventionalization of innovations at their various stages, lower-proficiency learner language and conversational spoken corpora are of essential importance. As regards identifying and choosing potentially interesting and relevant subjects of research, I hope that linguists (and journal editors) will remain brave enough to also direct their interest to features and research materials which are challenging for corpus searches and advanced statistics. A diversity of methods means a diversity of new information.