Lieselotte Anderwald set for herself a fascinating task: to examine how well metalinguistic descriptions of certain verb constructions in nineteenth-century grammars corresponded to the empirical record of those same verb constructions. We are used to seeing studies that document the empirical record, and Anderwald draws well on those studies and adds some of her own corpus searches. What is (much) less usual is the attempt to both generalize the body of metalinguistic commentary in grammars of the period and to relate that description and evaluation in those grammars with trends in the linguistic record. A few studies have included a detailed examination of the metalinguistic record (Visser Reference Visser1963; Dekeyser Reference Dekeyser1975; Yáñez-Bouza Reference Yáñez-Bouza2015), but for the most part, scholars like Anderwald are pioneering new territory, and part of the treat of reading a book like this is seeing new questions and new methods in evaluating diachronic change in English.
Anderwald employs a straightforward organization for her task. For the six (or so) structures that she examines, she first documents the historical distribution of the linguistic variants, such as the <t> vs <ed> endings for certain weak verbs. She then documents the comments on that structure found in the grammars of the eighteenth and nineteenth centuries, emphasizing the nineteenth-century grammars. She then compares the comments from the grammars with the historical distribution. Her last section for each construction is a recap of any insights we gain from that comparison. Such an organization gives Anderwald an opportunity to address the question head-on, and we are always reminded what the question of the book is.
Of course the sections on the historical distribution are easy enough to follow, since they rely largely on corpus studies. By now we are used to scholars trying to make the most of whatever corpora are available and accommodating the strengths and weaknesses of those corpora to the task at hand. To my eyes, Anderwald is skillful in her assessment of previous studies, where they are available. She weighs the adequacy of the corpus evidence and supplements previous studies with her own corpus searches, when she sees more insight to be gained from using larger corpora (particularly COHA) that are now available. The value of Anderwald's combining previous studies with her own corpus studies is shown, for example, when she discovers that the weak verbs with <t>/<ed> variation all had different histories, but that fact had been obscured when they had been grouped together in previous studies using smaller corpora.
It is Anderwald's next sections on summarizing grammars and relating them to the historical distribution that are more unusual for us to read, and, I would wager, harder for her to write. How does one generalize disparate comments in 258 grammars? As Anderwald demonstrates, evaluative comments in grammars can come in several forms, and sometimes even the absence of a comment might be a kind of evaluation. When a grammar lists the past-tense forms of to sing as ‘sang, sung’, for example, is the grammarian signaling that sang is preferable since it is listed first? And if the grammar lists only sang, does the grammar disapprove of sung? With commentary that is much more complex than simply ‘acceptable’ vs ‘unacceptable’, grammars present a challenge to anyone wishing to generalize from them.
One method for generalizing will be to categorize the comments so that they can be compared across grammars, and indeed, Anderwald mostly does just that. An advantage of this method is that we can start seeing some trends, as Anderwald's charts show. We can see, for example, that over time, grammarians became less critical of the progressive passive and that American grammarians were generally more critical than British grammarians (pp. 204–5). All in all, the insights that Anderwald garners by such a categorizing method are valuable and interesting. On the other hand, such a method ends up effacing particulars in grammars that may be important. Anderwald herself notes the limitations of categorizing with A Dictionary of English Normative Grammar 1700–1800 (Sundby et al. Reference Sundby, Bjørge and Haugland1993). Here we are left with the editors’ own evaluation of comments on eighteenth-century grammar, which might be sufficient for most circumstances, but which can be insufficient for a more fine-grained analysis. I am not sure that there is any way out of this problem: we will need some kind of categorizing to achieve generalizations, but we will then be at the mercy of whatever categorizing schemes and decisions were made. Again, I believe Anderwald was skillful in her work – she provides enough self-referential remarks to her categorizing process to give us confidence that she classified items as carefully as she could. Combing through grammars for their analysis and evaluation is tedious and time-consuming, and we should be grateful for Anderwald's painstaking work. Had she wanted to, Anderwald doubtless could have provided the more discursive descriptions of individual works that we often see in histories of linguistics (Michael Reference Michael1970; Finegan Reference Finegan1980; Baron Reference Baron1982; Law Reference Law2003; Yáñez-Bouza Reference Yáñez-Bouza2015), since she would have been very familiar with them. And while Anderwald often quotes particular grammars, her main method is to categorize the remarks so that we can have comparisons and quantification. I believe the trade-off works well for this book.
As straightforward as the question and organization are, the answers are not. That is another treat of Anderwald's book. By looking closely at the grammars and actual usage, Anderwald uncovers not just one overarching pattern, but many different patterns, each giving us additional insight into the relationship between grammar writing and language change. Essentially Anderwald provides six to twenty-seven case studies on changes in the verb system, depending on how we group her linguistic variables (do we count all the eleven strong verbs she examines as one case study or eleven, since each verb has a different history?). Because she compares so many constructions, Anderwald is able to uncover several insights into the intersection of grammatical commentary and actual usage. Those insights are what I find most interesting in the book.
On a fairly general level, Anderwald finds that there is no consistent relationship between language change and the evaluation of grammarians. We would expect that new usages would be criticized when they are noticed, but Anderwald found that this is not always the case. While the new passive progressive certainly was criticized, the new get passive was not criticized as much, and the evolving active progressive was generally praised. The insight that we may well not find a universal trend in the growth and decline of criticism is useful and unlikely to have come without several comparative empirical studies of this nature.
When it comes to the basic question of Anderwald's book – whether there is a causal relationship between prescription and language change – the author finds little conclusive evidence of causation, though she finds much good counterevidence: several changes occurred before any remarks were made in grammar or occurred irrespective of any remarks in grammar. She finds a few instances of possible causation, but they are not very conclusive, widespread or long-lasting.
Throughout the book several more insights emerge, coming from several directions. While Anderwald provides a summary of each chapter, enough insights are embedded within the chapter that one will be repaid for reading the entire book. For those interested in linguistic change, Anderwald provides a succinct summary of the research, complemented by her own corpus searches, for six verbal constructions listed in the chapter synopses below. One value of her book is that it collects scholarship on verb forms into one volume. This book can be a handy guide to the changes in verb forms throughout the nineteenth century – a century that has generally received less attention in the history of English, even though scholars have lately been showing that language was changing in that century as much as in any other. It is useful to have this reminder and demonstration in one place.
For those interested in how linguistic analysis in the past contrasts with today's analysis, Anderwald provides a couple of interesting gems, though such a contrast is not her major aim. For example, she shows how earlier grammarians saw the be perfect as a passive, since the construction was similar to the passive (be + Past Participle). Anderwald also shows that nineteenth-century grammarians often analyzed the passive progressive as an existential (not auxiliary) verb be, plus a participial construction (being + Past Participle). Anderwald provides several more instances of different linguistic analysis, especially in chapter 2.
My own interests gravitate to the workings of prescriptivism, and there are many insights about that. One, for example, is that the grammar writers varied considerably in their motives and justifications in their evaluation of the different verb constructions. A leading justification for objecting to a construction is what Anderwald calls the One Right Way Principle, following Zwicky (Reference Zwicky2009). The notion is similar to the isomorphic principle that one function should have one form, so that if two forms exist, there must be two functions. Grammarians frequently either tried to suppress variation to preserve this isomorphism or tried to distinguish more functions for the variant forms to specialize in.
Another basis for evaluation is the grammarian's own system of analysis. Anderwald points out that with the be perfect, many grammarians were so fixed on privileging form–function isomorphism that they assumed that the be perfect must be a passive, because it looked very much like the passive construction of transitive verbs. They noticed that the passive would not make sense with intransitive verbs, so the grammarians condemned the construction as illogical. This is a case of their own insufficient analysis getting in the way of accepting the construction. The same thing happened with the passive progressive. The last part of the verb phrase (being + Past Participle) looked very much like a (usually absolute) participial construction, in which the perfect aspect of the participle was felt to be more salient than the passive voice, as in this sentence from her book: ‘these arrangements being made, the travelers set out’ (p. 208). Because the progressive had an imperfect meaning, the grammarians could not see how a construction with a perfect meaning could be used within it. Consequently, the passive progressive was also condemned as illogical.
Anderwald delves a little into just what ‘logic’ must have meant to the nineteenth-century grammarians. Logic is often associated with other cultural terms which Anderwald distills as ‘A is logical, therefore it is beautiful’, ‘A is logical, therefore it is good’, ‘A is logical, therefore it is elegant’, and on the flip-side, ‘A is illogical, therefore it is ugly’ and ‘A is illogical, therefore it is morally reprehensible.’ Anderwald distills several other values, not all of them necessarily connected to logic. One is that ‘A is too complex, and therefore wrong.’ This is the basis of some objections to the passive progressive. Another value that Anderwald distills is that novelty may make a construction objectionable. This value starts out as ‘A is not used by the best writers’, but since the ‘best writers’ were usually writers from the past, the value can be reframed as ‘A is not used by seventeenth- and eighteenth-century writers’, which itself can be reduced to ‘A is new.’
Another basis of objection that Anderwald uncovers is the text-type that is most associated with a particular variant. Anderwald shows that some constructions seem to be sensitive to text-type, and it is on the basis of the associations with particular registers that the construction might be more criticized. Such is the case with the passive progressive. In her corpus analysis, Anderwald shows that the passive progressive showed up more often in newspapers than elsewhere. One objection to the passive progressive was that the wrong kind of writers –newspaper writers – were using the form. Anderwald quotes one of the nineteenth-century grammars to make this point: the passive progressive ‘occurs most frequently in newspaper paragraphs, and in hasty compositions’ (Weld 1848: 180; quoted on p. 201).
I found little to object to in this book. In general, Anderwald shows good instincts in her methods, and almost every time I could sense an objection arising within myself, Anderwald would answer it within a paragraph or two. I wonder a little, however, whether we would learn more about the relationship between prescriptivism and language change if we had some way of weighting those grammars that were the most influential. Anderwald's method gives us a welcome overview of the genre of grammars, as she quotes from a range of grammars. The averages of this range tell us a fair amount about how the constructions were viewed. On the other hand, if we are to see an effect from the grammars, won't the mechanism depend on which grammars were the most read or held the most prestige? This is not to say that Anderwald ought to have weighted the grammars in her study – only that such work remains to be done. In its own right, this is a welcome book full of data and insights that will be useful to scholars who continue to examine the prescriptive tradition in English or simply language change in English.
What follows is a chapter-by-chapter synopsis of the book.
Chapter 1 sets the research task for the book, namely to correlate meta-description with language change on a scale of decades. Such a task has rarely been done before, and Anderwald claims that scholars have been too ready to assign unexpected language change to prescriptive grammars. But instead of beginning with known prescriptions, Anderwald begins with constructions that were known to have changed in the nineteenth century and then looks to see what the commentary was. By beginning with the construction rather than the commentary, she is in a position to examine those constructions that elicit little notice or even receive approval. Anderwald also gives the details of her Collection of Nineteenth-century Grammars (CNG), which consists of 258 grammars, divided between the British Isles (mostly England) and North America (mostly the United States). She also gives details of the corpora she uses or refers to, particularly ARCHER, CONCE, COHA, Helsinki and BNC.
In chapter 2, Anderwald summarizes what the grammars say about verbs. She notes that different definitions of verbs depend on different criteria for defining (e.g. semantic vs syntactic criteria). This chapter looks particularly at the definitions of verbs, the classification of Active/Passive/Neuter that gave way to transitive/intransitive, regular vs irregular verbs, the nature of the future, and the status of the perfect. The discussion is useful and informative, but it is also the least integrated with the rest of the book. Some of the discussion is relevant to later chapters, but you could probably skip this chapter and come back to the relevant sections, if you like. I am not recommending that you skip it, however, since the metalinguistic description of verbs is interesting in itself.
Chapter 3 treats the variation in strong verbs. The main phenomenon that Anderwald treats is the variant past-tense forms for verbs like sing and sling. For the sing class, the variation is between <u> and <a> (sung vs sang). For the sling class, there is not much variation – <u> is almost invariantly used. Anderwald shows that the grammars sometimes indeed prescribed the a-form for both classes, and there was some little change two generations after the peak of such prescription. This suggests that there could have been an effect from the prescription. Anderwald further shows that British grammars seem to have been trying to introduce the a-form even for verbs that had been leveled to <u> over a century before (p. 80). The other subtopic Anderwald treats in chapter 3 is variation between strong and weak forms for certain verbs like dive, sneak and drag. She notes that changes occurred without much notice from the grammars, so the grammars do not account for the changes.
Chapter 4 treats changes to irregular weak verbs, like burn and dream. A notable contribution in this chapter is separating out the burn class from the dream class. For both classes of verbs, there is variation between the <t> and the <ed> spellings for the past-tense form, but verbs like dream additionally change the stem vowel, while verbs like burn do not. Previous studies had not made this distinction, but it turns out that the historical distribution is different for each class, so it is a distinction worth making. Anderwald also includes the verb to plead, which has variation in the vowel and ending (pled vs pleaded). Among the dream class, the <t> form became most prominent for kneel, but it remained a minority form for dream, leap, lean and plead, though there was a slight rise for leapt and pled in America. In Britain, we see a rise of the <t> form for dream, leap and lean in the second half of the nineteenth century. The change in leapt matches to a degree the trends in grammar that prefer leapt, so there might be a case of influence, but the overall use of the form is still so low (8%) that the grammars cannot have been a big factor. There may have been influence with plead in America, as well, since the recommendation of pled in grammars increased and so did the use in the corpora ten years later.
The burn class of verbs shows a decreasing use of <t> throughout the nineteenth century, except for dwell. It was only in the twentieth century that dwelled became the majority form. The preference for <ed> came after the change in actual usage, so it seems to be a case of the grammars catching up with usage.
Chapter 5 focuses on the variation between be and have as the auxiliary in the perfect tense/aspect for mutative verbs (e.g. I am arrived vs I have arrived). This variation had almost ended by the nineteenth century, but there were still a few instances of its use. Anderwald shows that the grammars analyzed the form in many different ways, and American grammars were slower than British grammars to recognize the perfect meaning of the construction. The be perfect largely disappears in writing over the course of the nineteenth century, yet grammars keep treating it. This illustrates the conservative nature of grammars.
Chapter 6 treats the active progressive, a form that was increasing in use throughout the nineteenth century. Grammar writers certainly noticed it, but instead of censuring it, they praised it as a useful construction unlike anything in French, German or Latin. The grammarians also labored to find a description of its meaning.
Chapter 7 treats two constructions: the passive progressive and the get passive, along with other get constructions (e.g. have got, get rid of). The first construction, the passive progressive, is the most criticized of the constructions that Anderwald investigates. Grammars find many reasons to dislike it (it is complex, it does not yield to their compositional attempts, it is new, it is used by the wrong sort of people), yet it increased in use during the nineteenth century. This is a case in which the proscriptions did not stop the language change. The get passive and other get constructions also increased in use, but they did not receive as much criticism as the passive progressive. Nonetheless, the peak of the criticism seems to have been in the 1830s, and about that time, there was a plateau in the use for about ten years, so perhaps the criticism does reflect an effect of prescriptive statements against a construction, but it is a weak effect. The use continued to increase after that plateau.
In chapter 8, Anderwald recaps several of the insights that we learn from the previous chapters. Some of the points she recaps are that grammars were often not very prescriptive, especially with regard to past-tense verbs. She also notes that the periphrastic constructions received more notice and perhaps more criticism, that the direction of the evaluation was not consistent and that the considerations that made one construction objectionable made another good. Anderwald notes some interesting questions to investigate, such as whether the speed of change and the type of feature undergoing change do play a role in prompting prescriptive statements. She also wonders if the phase of change (incipient, near the end) would also be a factor in the kind of criticism a construction receives. She brings up questions of frequency and register, as well. She points out that British grammars were less prescriptive and critical than American ones. She states that the impact of nineteenth-century grammar writing on actual language change has been greatly overestimated by previous scholars. She also includes a section on the underlying premises of prescriptive statements in the grammars.
The back matter of the book contains thirty-six pages of tables in an appendix, twenty-three pages of references to primary sources, thirteen pages of references to secondary sources, and a thirteen-page index.
