Diachronic linguistics has, from its beginning, been focused mostly on presenting theories of language change – largely because evidence, even in well-documented languages like English, has been notoriously difficult to find and/or collect. In this book, Dagmar Haumann and Kristin Killie seek to close this gap in research by building on previous work in the field of diachronic English adverb syntax and by using large-scale historical corpora to explore the development of sentence adverbs over the past 475 years. The English language has the benefit of significant textual evidence when compared to most of the world’s languages and is a strong candidate for such a study. In particular, this monograph uses a large dataset of fiction along with the frameworks of generative syntax and grammaticalization and examines twelve English adverbs (all with the -ly suffix) in the light of six existing hypotheses of language change. It is a complex task that the authors engage in, combining data and theory in new ways and, as such, this book is important to the fields of diachronic and generative linguistics.
The volume is comprised of nine chapters. In the first chapter, ‘Introduction’ (pp. 1–6), Haumann and Killie introduce their study by briefly explaining the multiple components that work together in their research. To begin with, this is a large-scale, quantitative analysis drawing on corpus linguistic methodologies; it is diachronic, so the corpora used must combine to cover a significant period of time. The authors limit the scope to three categories of speaker-oriented adverbs (SOAs): evaluative, evidential and illocutionary (or speech act). Haumann and Killie then contextualize their project in the tradition of diachronic adverb research going back to Swan’s (Reference Swan1988) foundational work. Moving on to the uniqueness of their study, they briefly introduce the six hypotheses, pulled from several sources, which they will test with corpus data: three reanalysis hypotheses, two lexicalization hypotheses and an analogy hypothesis. The three selected corpora covering the period from 1550 to 1899 in English include Early English Prose Fiction, Eighteenth-Century Fiction and Nineteenth-Century Fiction; the robust use of corpus data to supplement qualitative example sentences helps to set this study apart. The chapter closes with an outline of the book’s structure.
In chapter 2, ‘The syntax of SOAs and their narrow-scope counterparts’ (pp. 7–45), Haumann and Killie focus on the task of connecting their research to generative linguistics frameworks, including minimalist phase theory (e.g. Chomsky Reference Chomsky and Kenstowicz2001) and syntactic cartography (e.g. Rizzi Reference Rizzi and Haegeman1997; Cinque Reference Cinque1999). Formal approaches to diachronic syntax are still relatively recent, and the authors are careful to explain (and show through the use of figures) the ways in which the minimalist and cartographic frameworks together are suitable for the study of subject-oriented adverbs. These two frameworks have not always been considered compatible, even from within a formal approach; however, Haumann and Killie argue for the reconciliation of supposed oppositions between the two models.
Using Cinque’s (Reference Cinque1999) cartographic structure, strictly laying out the placement of adverbs throughout the sentence, Haumann and Killie elaborate the Left Periphery and the medial structures of the sentence, focusing on the Polarity Phrase (PolP; home to verbal negation in English). The PolP is then proposed to be the site marking the boundary between the generative vP/VP phase (the initial phase of sentence generation) and the CP/TP phase (the final phase), largely because it seems to be a boundary between narrow-scope (vP/VP) adverbs and subject-oriented (CP/TP) adverbs. As a result, it is also a likely locus for structural ambiguity, which is crucial to the hypotheses of syntactic and semantic reanalysis of adverbs in the history of English. This chapter, with its thorough situating of diachronic adverb studies within generative theoretical frameworks, will be something of a challenge to some scholars, as the generative–diachronic crossover community remains a minority among both formal and historical research approaches.
Haumann and Killie continue the work of connecting generative and diachronic studies in chapter 3, ‘Theoretical underpinnings: The role of reanalysis and analogy’ (pp. 46–65). Here, they examine various aspects of grammaticalization theory, which is more familiar territory for most diachronic linguists. They define grammaticalization as an abrupt change which involves syntactic reanalysis and semantic reinterpretation, frequently related to structural ambiguity. Well-studied examples such as English be going to and French ne…pas are presented, and the authors then explore the contrasts between primary and secondary grammaticalization, on the one hand, and between grammaticalization as reduction and as expansion, on the other. In this study, Haumann and Killie focus on secondary grammaticalization and grammaticalization as expansion, as adverbs in -ly are expanding to new contexts and new meanings within the sentence.
The authors view subjectification and intersubjectification as typical semantic and pragmatic markers of grammaticalization, noting that all of the adverbs under consideration here demonstrate these features. Part of the (inter)subjectification that is important to the process of grammaticalization includes what Traugott (Reference Traugott, Allan and Jaszczolt2012) calls invited inferencing or implicatures. Haumann and Killie discuss the fact that such inferencing may have a different effect than intended by the speaker/writer, and these so-called bridging contexts offer the type of ambiguity needed for reanalysis. Such bridging contexts should then be followed by switch contexts, showing the reanalyzed lexical item in its new, unambiguous use. After carefully exploring the parameters of grammaticalization by reanalysis, the authors turn briefly to questions of reanalysis without ambiguity and questions of analogy. Analogy in particular is discussed as an unclear concept that may refer to a process, its product or even a cognitive faculty. As a result, the authors are wary of its use in an analytical way, while admitting that it is nevertheless used by many scholars.
In the fourth chapter, ‘Hypotheses about the development of speaker-oriented adverbs in English’ (pp. 66–83), Haumann and Killie present the six hypotheses postulated in this study. Before doing so, however, they return to Swan, whose scholarship in the late 1980s and early 1990s has inspired their own work and the work of the other researchers on which their approach draws. The relative lack of examples from Old English (largely limited to OE truth conditionals soþlice and witodlice) until the Modern English period is first noted by Swan, and much of her data continues to feature in present-day discussions. The authors move from Swan, whose work they consider untestable, to what they call Hypothesis I: that SOAs arise from narrow-scope adverbs through reanalysis. This first hypothesis would be strengthened by chronology; narrow-scope adverbs should precede SOAs, as should ambiguous cases. Assuming Hypothesis I is adopted, there are two related hypotheses. Hypothesis II assumes reanalysis in the Left Periphery or initial position of the sentence, while Hypothesis III proposes reanalysis in the medial position (between the subject and the lexical verb). Hypothesis II would be strengthened by a large number of ambiguous adverbs in initial position, and Hypothesis III would be strengthened by a large number of ambiguous adverbs in medial position.
After discussing the reanalysis hypotheses, Haumann and Killie move on to others. They first discuss ellipsis accounts, settling on a lexicalization-by-ellipsis hypothesis (Hypothesis IV) that uses It is Adj that structures as the source for SOAs (specifically evidential and evaluative), with the -ly morpheme acting in place of a copula. This hypothesis would be strengthened by a large number of It is Adj that structures in the corpora. Hypothesis V is another lexicalization account; this hypothesis is for illocutionary adverbs, which arise in external clauses that are then connected to the main clause, also through ellipsis. It would be strengthened by examples of external and subordinated clause structures preceding illocutionary SOAs. The final hypothesis (VI) explored by the authors is focused on analogy. For Haumann and Killie, this is essentially the consideration of analogues rather than of a cognitive process of analogy. As they note, some SOAs are not preceded by narrow-scope adverbs and thereby fall out of the reanalysis or lexicalization hypotheses altogether. Chapters 6–8 will consider these six hypotheses in light of the corpus data the authors have gathered.
Chapter 5, ‘Corpus, data and method’ (pp. 84–96), focuses on the details of Haumann and Killie’s empirical evidence. Based on previous studies, including the early work by Swan, the authors choose the year 1550 as the starting point for their analysis. This date is early enough to capture all of middle-to-late Modern English. They decide not to focus on Present-day English, so the corpus ends in 1899. The three historical literature corpora examined (Early English Prose Fiction, Eighteenth-Century Fiction, Nineteenth-Century Fiction) have a total of 59 million words, which is a large enough number for a strong quantitative case. After discussing the corpora, Haumann and Killie move to the adverbs under scrutiny. Among illocutionary adverbs, they distinguish between content-related illocutionary adverbs (frankly, honestly, seriously), which comment on the content of the proposition, and form-related illocutionary adverbs (briefly, broadly, roughly), which comment on the proposition’s form. Additionally, they analyze three evaluative adverbs (fortunately, happily, surprisingly) and three evidential adverbs (apparently, evidently, obviously). Their choices are based on the number of tokens in the corpora and on the adverbs that have been discussed before in the literature; they have also limited the scope to adverbs ending in the -ly morpheme. There are 13,164 tokens for these twelve adverbs, and 12,891 tokens that are either narrow-scope or subject-oriented adverbs. The adverbs are organized into groups based on fifty-year intervals and are labeled as narrow-scope, as SOA or as ambiguous. The tokens are classified as well as part of finite or non-finite clauses, as part of matrix or subordinate clauses, and as being in initial or medial position. If there is more than one adverb in a clause (including negation), the linear order is indicated, and any adverb set off by commas is also annotated.
With chapter 6, ‘Illocutionary adverbs’ (pp. 97–120), Haumann and Killie begin to discuss their corpus data and offer analyses. They start by defining and giving examples of the six illocutionary adverbs. The authors then analyze the 4,749 tokens of illocutionary adverbs, testing the hypotheses separately. The data are not conclusive for the reanalysis hypotheses (I–III): while there are many more narrow-scope examples throughout the corpus than SOAs, there is almost no ambiguity observed. Haumann and Killie therefore reject these three hypotheses. Hypothesis IV, lexicalization through ellipsis, has very little empirical support as well, with few examples of any stage of that hypothesis’s multi-step process. Hypothesis V does not apply, so the authors move on to Hypothesis VI, analogy. Analogy seems to be possible for illocutionary adverbs, although it serves primarily as an ‘elsewhere’ condition. In light of limited results, the authors offer an important alternative proposal: the collocation of illocutionary adverbs with verbs of communication. Although this is similar in some ways to Hypothesis IV, it is less specific and is more fully supported by the data. All illocutionary adverbs collocate frequently (though at rather different rates) with communication verbs, and the authors propose a semantic ‘absorption’ through ellipsis of the verbal element onto the adverb, which is then licensed to act as an illocutionary SOA.
Chapter 7, ‘Evaluative adverbs’ (pp. 121–41), continues the same pattern as the previous chapter. Haumann and Killie begin by defining the three evaluative adverbs and then examine the total tokens (2,803). More than two-thirds of the tokens belong to happily, while surprisingly only comprises 3.6 percent. Across the time range of the corpus, the number of narrow-scope examples consistently decreases, while the number of subject-oriented adverbs increases. The adverb fortunately shows the greatest shift, while happily is similarly affected. The limited data for surprisingly does not allow for generalizations. Hypotheses I and III are supported at least to some extent, as there is more ambiguity in medial positions than in the initial slots. Hypothesis V (It is Adj that structure) is not supported because of a lack of data, similar to Hypothesis IV for illocutionary adverbs. Analogy, or Hypothesis VI, is unclear.
In chapter 8, ‘Evidential adverbs’ (pp. 142–55), Haumann and Killie complete the pattern used across the data chapters. In the first subsection, however, they pause to distinguish evidentiality from epistemicity for the purposes of their study. The authors discuss the advantages of different taxonomies of evidentiality while explaining that many languages often specify evidentiality through the grammar. English does not, and as a result evidentiality can overlap with epistemicity by way of modal auxiliaries, adjectives and adverbs. Some modals, for example, demonstrate both features. The selected adverbs (apparently, evidently, obviously), however, are primarily evidential. After clarifying this, the authors define the words and discuss their occurrence in the corpus. There is a total of 5,339 tokens, with evidently (2,630) and apparently (2,386) dividing the majority of them. Although the temporal shift from narrow-scope to SOA is strong across the board, the authors contend that Hypotheses I–III are not supported because there is essentially no ambiguity. With regard to Hypothesis V, the negligible number of It is Adj that cases for apparently and obviously plus the fact that a larger number of cases occur for evidently (but are coincidental with the SOA use) combine to account for its rejection by the authors. In light of these considerations, analogy (Hypothesis VI) seems fairly likely.
Chapter 9, ‘Concluding discussion’ (pp. 156–61), allows Haumann and Killie to bring everything together, but the results are admittedly mixed. Hypothesis I is supported to an extent in some instances, but overall the authors point to a weakened version as likely, namely that SOAs may but need not develop through reanalysis. They find no support for the hypothesis that the reanalysis (when it occurs) happens in initial position but some support – especially for the evaluative subset – for it happening in medial position. The lexicalization hypotheses (IV and V) are rejected because of minimal data in most cases and a problematic chronology for the evidently case. Analogy (VI) is a likely mechanism that cannot be proved. Ultimately, they suggest a multiple-source account for the development of SOAs. Illocutionary adverbs may result from a collocation of manner adverbs and communication verbs. Evaluative adverbs may result from the reanalysis of ambiguous adverbs in medial positions. And evidential adverbs may result from analogy.
After discussing the results, Haumann and Killie express some thoughts about the future of the field. First, the variability of the data is surprising. The lack of structural ambiguity in the corpora is unexpected, considering that all scholars support a grammaticalization account that largely relies on ambiguity. In addition, the patterning is highly idiosyncratic across adverbs (both types and individual lexical items). Because of these concerns, Haumann and Killie advocate for studying more adverbs across different text types and genres, with a greater concentration on twentieth-century corpora.
Despite the mixed results, this study stands out for several reasons. It is the first book-length study of subject-oriented adverbs in many years, and it contributes to the understanding and scholarship in the field. In addition, it is the largest empirical study of SOAs, with a significant number of tokens and adverbs studied. It offers important contributions to the field of generative syntax, with the focus on the Polarity Phrase as a phase boundary, and to the field of grammaticalization studies, with the comparison of reanalysis, lexicalization and analogy. The authors have a facility with the hypotheses and approaches found in the varied literature about adverbs, and their discussions are clear, thoughtful and highly detailed. This study provides researchers with a new starting point and new directions for future projects, and it is a welcome addition to the library of texts in diachronic, generative and corpus-based linguistics.
