Rautionaho Paula, Arja Nurmi and Juhani Klemola: Corpora and the changing society: Studies in the evolution of English

Corpus linguistic methods have become a mainstay of historical sociolinguistics, and the present volume attempts to reflect the state-of-the-art in this field. The intention of the editors is to showcase exactly how corpus linguistics can be used to see how social change is reflected in language change (both in terms of short- and long-term timeframes), and the 11 papers here all emerge from the ICAME 2018 conference. The volume is divided into two sections: the first one (“Changing society”) focuses on studies examining broader social change as reflected in language change, while the second one (“Changing language”) features in-depth studies of particular linguistic items.

After a brief introduction by the editors, Martin Hilpert’s contribution, “The great temptation: What diachronic corpora do and do not reveal about social change”, opens the volume proper. Hilpert discusses the benefits and disadvantages of using large datasets to study language change and presents us with five possible pitfalls in the use of such datasets (similar to Rissanen’s 1989 discussion of general challenges faced in using diachronic corpora): semasiological frequencies are not necessarily equivalent to similar changes in onomasiological frequencies; frequencies of polysemous words need further distinguishing along sense- and construction-specific uses; correlations in large datasets may be spurious; frequency trends found in (large) diachronic corpora require adequate statistical treatment; and linguistic change is not always easy to disentangle from social change. Hilpert then illustrates how these challenges can be confronted in a case study of the development of the make-causative construction (e.g. Sally made Ben cry) using data from COHA (Corpus of Historical American English).

Gerold Schneider focuses specifically on the social phenomenon of poverty in the next chapter, “Changes in society and language: Charting poverty”, using data from EEBO (Early English Books Online) and CLMET 3.0 (Corpus of Late Modern English). Schneider employs methods associated with document classification, topic modelling and distributional semantics, and kernel density information to see how the notion of poverty can be charted in the history of English. He finds topic modelling and kernel density estimation to be of most use here in accounting for the conceptualisation of poverty alongside a number of other social matters (religion, trade, war), and notes that it is critical that data interpretation must contain rigorous qualitative analysis beyond number crunching.

Medicine is the focus of the next contribution, in which Maura Ratia discusses the process of “Finding evidence for a changing society: A collocational study of medical discourse in 1500–1800”. Numerous advances in medicine were made during the three centuries that Ratia investigates, and she is particularly interested in the view/role of the patient during this period. Ratia uses the EMEMT (Early Modern English Medical Texts) and LMEMT (Late Modern English Medical Texts) corpora to trace the collocations, semantic prosody and lexical bundles exhibited by the lexeme patient. She finds, among other things, that the focus in EMEMT is often on treatments and the role of physicians (with the patient mostly in object position), whereas in LMEMT the emphasis shifts to the patient’s experience and feelings of well-being (with the patient often in subject position).

Antoinette Renouf is particularly interested in the study of neosemes (new meanings attached to existing words) as possible reflectors of social change in her contribution, “Semantic neology: Challenges in matching corpus-based semantic change to real-world change”. She deploys the WebCorpLSE’s automated collocation analysis software package to study a corpus of newspaper texts from 1984 to 2017 in a number of smaller cases studies: birther (a conspiracy theory focusing on the location of former President Obama’s birth); normalisation (of certain types of political rhetoric and behaviour); cougar (an older woman involved with a younger man); snowflake (purportedly hypersensitive members of the younger generation with a sense of entitlement); and ghosting (the abrupt termination of a relationship by cutting off communication). Collocational profiles are examined in detail and temporal developments are traced with time graphs. Renouf finds the methods used fairly successful, although she notes steps can be taken to improve system performance.

Gavin Brooks and David Wright’s contribution, “From burden to threat: A diachronic study of language ideology and migrant representation in the British press”, concludes the first section. Here, the authors examine the collocational behaviour of the phrase speak English in a corpus of right-wing press articles (collected via Nexis) from 2005 to 2017 (with the 2011 census serving as the central point). Four themes are examined in detail: proficiency, multilingualism, learning English and integration, and public/private services. The authors combine their quantitative results with qualitative explanations responsible for any apparent change (e.g. the right-wing press’ focus on the detrimental effects of non-proficiency on the national health service without regard to the increasing effects of austerity on the organisation).

Part Two (on “Changing language”) opens with “That’s absolutely fine: An investigation of absolutely in the spoken BNC2014”, in which Karin Aijmer traces the evolution of absolutely from an intensifier to a pragmatic marker of subjective assessment. She notes some developments between 1994 and 2014 along the lines of age and gender (older speakers tend to use the marker more often; females have outpaced males in usage, the latter having constituted the majority of users in 1994). The role of syntactic headwords, as well as the on-going role of grammaticalisation and subjectification in the use of absolutely, is also given some attention.

Zeltia Blanco-Suárez’s contribution examining “Two sides of the same coin: Tracking the history of the intensifiers deadly and mortal” is next. Blanco-Suárez uses a number of diachronic corpora to trace the grammaticalisation and subjectification of deadly and mortal from Old English to the present day (highlighting the development of descriptive > affective > intensifying meanings). She finds that affective meanings become more common as time passes, and intensifying meanings do not even emerge until the early modern period. Deadly is most often accompanied with negative shading, whereas mortal exhibits a greater variety of positive/negative/neutral shadings.

Yoko Iyeiri focuses on “So-called -ingly adverbs in Late Middle and Early Modern English” in her discussion, whereby larger datasets such as EEBO are used to test earlier studies, which did not have such resources at their disposal. Her examination confirms the previous assertion that such adverbs rose drastically during the Late Modern English period, but that this development is more nuanced than previously believed (due to the types and range of adverbs used). Iyeiri also examines the use of “Harry Potter” -ingly adverbs (used frequently in the work of J. K. Rowling and often involving verbs of saying, motion and watching; e.g. He shook his head unsmilingly) and notes their possible emergence during the early modern period, in contrast to the previous claim that they emerged during the late modern period.

Martin Schweinberger combines the methods of frequency analysis, semantic vector space and collexeme analyses in his discussion of “Analyzing change in the American English amplifier system in the fiction genre”, whereby the COHA corpus is consulted to trace the development of amplifiers in fiction from 1850 to 2000. Schweinberger finds that the amplifier so is outpacing very in terms of frequency, but there is no broadening in its usage or increased lexical diversity in terms of its collocational profile. Comparisons are made with the behaviour of amplifiers in other varieties of English.

Laurel J. Brinton examines “The development and pragmatic function of a non-inference marker: That is not to say (that)” in her contribution, in which she argues for a discourse-pragmatic reading of this phrase. This is based on her tracing the grammaticalisation of this phrase through a number of diachronic corpora (going back to EEBO and Early Modern English). She notes the difficulty in tracing its development due to the prevalence of fiction in several corpora, whereas modern corpus data reveal this construction’s usage is highest in non-fiction written prose. Attention to the presence/absence of the complementiser that is discussed, and Brinton notes that since this construction is already attested in the sixteenth century, studies going back even earlier into Middle English are necessary for a full diachronic picture of this construction to be captured.

Turo Vartiainen and Mikko Höglund’s discussion of “Changes in transitivity and reflexive uses of sit (me/myself down) in Early and Late Modern English” wrap-up the volume. The authors use the Transitivity Hypothesis (first introduced in Hopper and Thompson 1980) to explore the evolution of the reflexive construction involving both the simple (me) and reflexive (myself) pronoun strategies using a number of large diachronic corpora (similar to Iyeiri’s study, earlier studies on this topic made use of comparatively small datasets). They note that these two constructions initially occupied different functional spaces (with the simple strategy often associated with sit down) and that the role of telic vs. atelic aspect has played a significant role in the development of this construction. Confounding the data are emphatic uses of reflexive pronouns, which were not of immediate interest to the authors. Vartiainen and Höglund note the applicability of the Transitivity Hypothesis to the study of diachronic developments with their study.

Overall, the volume Corpora and the Changing Society offers some welcome contributions to the fields of historical corpus linguistics and historical sociolinguistics. One of the key and most significant emphases of this volume – made explicit in Part One and implicit throughout Part Two – is the indispensability of close reading and detailed qualitative analysis alongside the more well-known methods of distant reading and quantitative analysis associated with corpus approaches to large datasets. Hilpert’s contribution is a standout here, as he offers the most wide-ranging and impactful analysis through his discussion of common pitfalls associated with working with large datasets: something that anyone working with “big data”, whether through a present-day or historical lens, would do well to heed. All other essays in Part One also demonstrate this, albeit in more focused and topic-specific ways. And they all illustrate exactly how fruitful the use of corpus linguistic methods can be in tracing the developments of social change.

That being said, the focus of the essays in Part One alongside the editors’ general introduction about the connection between language change and social change make the contributions of Part Two seem a bit out of place here. The whole premise of the volume is purportedly to show how corpus linguistic methods can help us trace broader social changes alongside linguistic ones. Yet the contributions in Part Two, while successfully using such methods to track the development of specific linguistic items (often within the context of grammaticalisation and subjectification), say little if anything about the broader social implications of such change, or at least how such change reflects broader sociohistorical developments.

All the contributions to the volume are solid, well-researched and thought-out discussions of the topics at hand, but there are a few oversights that the authors and editors should have picked up on. Renouf’s discussion of neosemes is very interesting and her methodological approach is certainly insightful, although one of her time graphs (Figure 2 on normalisation, p. 88) does not distinguish between the different senses of the word (unlike the other figures), and the significance of the information is thus left imprecise. Blanco-Suárez’s contribution on intensifiers demonstrates the possibility of tracing long-term diachronic developments using a number of different corpora; however, a discussion of the implications of using datasets of vastly different sizes on the treatment of quantitative information was completely absent. The key to one of her figures (Figure 3, p. 189) appears to be incorrect, although one can simply consult Figure 4 immediately below to see what the visualisation actually shows. And I would have liked to see Iyeiri devote more attention to the “Harry Potter” adverbs in Early Modern English in her contribution, making distinctions between the “objective” and “subjective” uses she found; this would have allowed her contribution to complement the broader theme of subjectification found in Part Two of the volume more explicitly.

These observations aside, this volume successfully serves as a window into the state-of-the-art of historical corpus linguistics. All contributions demonstrate a meticulous blend of quantitative and qualitative analyses with conclusions being well-grounded and explicated thoroughly. The essays in Part One also make valuable contributions to the field of historical sociolinguistics, whereas those in Part Two offer more nuanced discussions of the evolution of particular linguistic items, oftentimes (but not exclusively) discussed within the frameworks of grammaticalisation and subjectification.