Language

Anja Wanner's research monograph on English passive constructions fills a gap in the longstanding discussion of passivization by bridging the divide between formalist and functionalist approaches and by focusing on syntactic, semantic, and pragmatic issues on a wide and well-informed basis.

The monograph comprises six chapters, partly based on empirical research findings emanating from corpus studies. Starting out with an introduction (Ch. 1), W poses the question of what motivates speakers' choice of the passive over the active, given that the passive (i) is more marked morphologically, (ii) deviates from the prototypical link between position and semantic role in which subjects are agents and objects are themes, and (iii) is generally harder to process than its active counterpart. An easy-to-read theoretical overview of the passive is provided in Ch. 2. In Ch. 3 the ingredients of the passive are described and empirically analyzed on the basis of the Freiburg-Brown (FROWN) corpus—with a laudable twenty-page extension dedicated to the GET-passive. Ch. 4 is reserved for a detailed treatment of the agent in passives, leading up to the core section of the book. It presents a synchronic corpus analysis of the effects of genre differences, which is related to previous findings on the diachronic development of the passive. The conclusion (Ch. 6) sums up the main arguments and discusses their implications in a concise and highly readable fashion.

The book begins with a sound introduction to the state of the art, addressing the relevance of analyzing passives for comparative linguistics, generative frameworks, and functional and discourse-oriented [End Page 452] research. When addressing theoretical issues, W successfully tries to bridge the gap between formalist and functionalist approaches. Though some assumptions formulated in principles-and-parameters approaches and restated throughout the book may not be uncritically accepted by all (e.g. claims about a movement operation involving a 'rearrangement' of major constituents (31, 53)), W convincingly shows that by making the perspectives of formalist and functional approaches visible to each other, we gain new insights into the operation of the English passive. Along with Comrie (1988) she holds that the active can be considered basic in terms of four parameters: (i) frequency (the active is more frequent than passive), (ii) formal complexity (the passive has more morphemes), (iii) productivity (all verbs used in the active can be used in the passive, but not vice versa), and (iv) discourse distribution (the active is preferred unless there are discourse-based reasons to use the passive). Ch. 2, 'The English passive and linguistic theory', summarizes previous formalist and functionalist research on passive constructions, as well as developing a morpheme-based definition of the passive. W follows Haspelmath's (1990:25) suggestion that the passive is foremost a verbal morphological category, since passive constructions without a passive participle do not exist, whereas passives do occur (i) without theme subjects (It was assumed that ...)—that is, there is no 'movement' of an object NP to subject position; (ii) without by-phrases—in fact, these cases are far more frequent than long passives (cf. Biber et al. 1999:938); and (iii) without BE (She got promoted) or as bare passives without any other explicit auxiliary (The race run by Harry (15)).

Rather than defining the passive on the basis of the traditionally employed syntactic template (NP BE Ven by NP), W opts for a wider-scope definition that is able to encompass a whole range of functionally overlapping passive constructions that cannot be accommodated in more traditional conceptions. Her definition of passives requires a passive participle, an implicit external argument that is not in subject position, and a propositional content equivalent to that of its corresponding active sentence. This wider definition also permits her to account for the contested status of GET as an auxiliary by acknowledging that negated sentences containing GET still require do-support: She was/*got not promoted (17). Thus, the morpheme-based definition proves better suited to encompass highly divergent passive constructions, such as raising structures (He was expected to leave), pseudo-passives (This bed has been slept in by Queen Victoria), recipient passives (He was given the letter by me), and GET-passives (My license got taken away). By contrast, it excludes nominal passives (The architect's construction of the concert hall), middle constructions (The shirt irons easily), and adjectival passives (She felt disappointed), for instance. Providing a general account of several active-passive asymmetries, W holds that the addition of the participle during passivization causes asymmetries in which a passive verb can, for instance, lose its capability to license a direct object but not its capability to take a clausal complement.

1. 1.

   1. a. *They explained the problem.
      *It was explained the problem (by them).

   2. b. *They explained why we should leave early.
      *It was explained (by them) why we should leave early.

(16)

This contrast is related to the case filter that assigns case to NPs but not clauses.

Drawing on pertinent work by Dowty (1991), Hopper and Thompson (1980), and Tenny (1994), W expertly discusses the relation between passivization and transitivity, concluding that 'passivization decreases the structural, semantic, and aspectual asymmetry between the participants' (37).

Ch. 3, 'Ingredients of the English passive', contrasts lexicalist (i.e. verb-based) and construction grammar approaches to English passives, arguing that the former are better suited to account for a whole range of formally divergent but functionally equivalent passive variants. While a construction grammar treatment requires establishing a whole set of different constructions, thus missing out on the semantic similarities shared by these constructions, the verb-based approach, according to W, permits us to treat the passive participle as the common denominator that largely determines syntactic and semantic properties of the passive. W also provides a highly readable outline of generativist principles adduced in order to explain why some verbs can occur in the passive while others cannot. The chapter closes with a corpus-based analysis of the GET-passive, [End Page 453] adding further support to the observation that passives are not restricted to formal style or written registers.

Ch. 4, 'The implicit agent in English passives', addresses the implicit argument of passive constructions. Based on the assumption that the common denominator of the various types of passives is the presence of the passive participle and the existence of an implicit argument, W discusses theoretical implications of this analysis. Moreover, she argues that typical properties of GET-passives—such as the responsibility reading for the nonagent subject (John got fired vs. John was fired) and portraying events as nonbeneficial for the subject (get beaten up)—derive from the semantic properties of GET rather than the passive construction.

Ch. 5, 'The use of the passive in academic discourse: A case study', provides several corpus studies—partly previously published in Dorgeloh & Wanner 2003, 2009. Following a quite entertaining account of prescriptive comments in usage guides, W conducts quantitative and qualitative corpus-based analyses of the passive in the language of academia, the 'natural habitat' (4) of the passive. This study is based on her own compilation of an abstract corpus (ca. 22,000 words), a section of the FROWN Corpus (ca. 160,000 words), and diachronic data from the science sections of A Representative Corpus of Historical English Registers (the ARCHER corpus) (ca. 140,000 words). At roughly seventeen cases per thousand words, the ratio of passives in the synchronic corpora parallels that observed by Biber and colleagues (1999). Every fifth passive in the abstract corpus came without an auxiliary, for example, in participial relative clauses (e.g. references overlooked in the previous compilations) and participial adverbial clauses (e.g. As pointed out in the previous section) (171). Many corpus-based studies on the English passive have neglected or straightforwardly excluded these passives, restricting their analysis to finite clauses, mostly for practical reasons relating to the time-consuming retrieval of participial passives from electronic corpora. It is one of the merits of the present book that the inclusion or exclusion of passives is based on clear-cut linguistic criteria.

W's genre-specific analysis focuses on those parts of scientific abstracts in which an author constructs an argument, so-called 'reporting events', and she contrasts these with 'reported events', in which the author summarizes work by others, for example. W mentions that her reporting events match Swales's (1990) 'commentaries', but she does not inform the reader about why she chooses to coin a new term. The data reveal that reporting events are less frequently found in journals on natural science than they are in the humanities. And within the 'reporting events' sections the passive is less frequently used in the humanities than in experiment- or observation-based journals. Remarkably, the low use of passives is not compensated for by rising numbers of explicit agents in actives, but by a third alternative that combines active voice with nonagentive subjects: W brands this type the 'paper-construction' (This paper rejects the idea...) and the 'fact-construction' (The data demonstrate that...). Both are highly frequent in scientific abstracts. W conceives of them as less agentive than even the passive, since in the passive 'the agent of the event is an implicit argu-ment' while in the paper- and fact-constructions 'the agent slot is filled with a non-agent argument, which gives less visibility to the actual agent than the passive does' (183). She thus establishes a cline of visibility of the agent from active via passive to fact-/paper-constructions. One might question whether the fact-/paper-constructions should be ranked as less agentive than the passive. We would require independent support showing that the author in a passive (It will be shown that ...) is cognitively more present than a nonagentive paper-construction (This paper will show that ...). What remains, however, is the remarkable frequency of the fact-/paper-construction. As regards the functional spectrum of passive constructions, the main competitors of passives are indeed actives, but not those that make the agent any more explicit than their passive construction. W concludes that the functional motivation for using the passive can thus—at least in the majority of cases—not be the often-stated desire to background the agent (192).

W also raises the issue of face, since 'reporting events' often contain material criticizing other researchers. Both passives and the fact-/paper-construction can be used to avoid explicit mention of the author as the source of a face-threatening act.

Ch. 5 closes with a diachronic study of eight verbs in the ARCHER corpus (argue, demonstrate, indicate, show, suggest, examine, explore, find) that allow active and passive constructions. [End Page 454] It is not clear on what basis these verbs have been singled out (e.g. frequency, ease of retrieval), or whether they are considered representative of other functionally similar verbs, such as report and investigate. But we clearly learn that, in general, the frequency of these eight verbs has increased over time. What is more, the ratio of passives used with these verbs has also generally increased, with two—unexplained—troughs in the second halves of the eighteenth and late twentieth centuries. The mention of the agent has decreased, particularly in the late nineteenth and throughout the twentieth centuries.

A highly consistent trend is the remarkable increase of the fact-construction in the time-spans investigated. W observes that in the early twentieth century the paper-construction appears on the scene. This observation is in line with another research result not yet available at the time of publication of W's book: a more general increase of inanimate subjects in passives from Old English to present-day English observed by Toyota (2008:161).

Throughout the book, W successfully challenges claims about the alleged ungrammaticality of certain constructions by referring to corpus data or simply by Googling. Thus, she counters Huddleston and Pullum's (2002:961) argumentation that that-clauses after raising verbs and after passivized verbs cannot be coordinated by providing convincing counterexamples: it ... was often other than it seemed, and was believed, to be.

The introduction ambitiously announces its aim to 'analyze whether recommendations found in style sheets and writing manuals have any effect on the use of the passive' (4). W observes that the passive has been increasingly used for centuries, but has recently declined. Similar to Seoane 2006 and Mair & Leech 2006, W's data show a drastic drop in the use of passives in the second half of the twentieth century. During this decline of the passive, a third alternative (This paper argues ...) has been gaining ground: 'This complementary pattern gives rise to the interpretation that the increase of the paper-construction is, to some extent, motivated by an effort to avoid the passive' (201). While this is a striking correlation that coincides with a prescriptive stigmatization of passives, the mere existence of the correlation does not necessarily imply causality. Whether the decline in passive constructions in the twentieth century is really caused by prescriptivism cannot yet be assessed.

While the asymmetries in active-passive sentence pairs with regard to the occurrence or nonoccurrence of the by-phrase and the question of which verbs allow passivization are frequently addressed in the literature and also in the present book, other asymmetries have gone largely unnoticed in W's book. They concern two only seemingly opposing observations. First, passives generally tend to opt for more explicit variants than the corresponding actives (cf. Andersson 1985:63-65, Rohdenburg 2000:34, 35, 2013, and Rohdenburg & Schlüter 2009 for notable exceptions). Thus, when language users are faced with the option of choosing between more explicit (e.g. to-infinitive, upon) versus less explicit variants (e.g. bare infinitive, on), the more explicit variant is comparatively preferred in passives rather than actives.

1. 2.

   1. a. Mum and Dad helped her Ø foot the bill.

   2. b. She was helped to foot the bill by Mum and Dad. (based on Rohdenburg 2013, ex.3)

2. 3.

   1. a. Jeanne and his other children prevailed on him to accept.

   2. b. ... he was prevailed upon to accept an appointment on the Committee on Social Thought at the University of Chicago,...

(COCA)¹

These observations can be related to W's argumentation that 'passives are costly' in terms of processing (37). In line with the complexity principle, which states, 'in the case of more or less explicit grammatical variants, the more explicit one(s) tend to be favoured in cognitively more demanding environments' (Rohdenburg 1996:151), we would expect the more explicit variant (e.g. upon) to be favored over its less explicit variant (e.g. on) in passive constructions.

Strikingly, a seemingly reverse trend has been observed for some constructions. A largely prescriptive early publication mentions the use of the less explicit variant in the following example: [End Page 455] '... "the terms have now been agreed" [instead of agreed to—BM] is less objectionable—perhaps because the preposition would come at the end, and since it is an insignificant word its absence is not noticeable' (Wood 1962:11, cited in Rohdenburg 2013). And indeed, corpus studies show that when faced with the option of choosing between prepositional and direct objects in the active and passive, the zero option unexpectedly turns out to be more frequent in the passive rather than in the active (cf. Rohdenburg 2013).

1. 4.

   1. a. An extensive body of literature also attests to unusually close and positive relations between foreclosed adolescents and their parents.

   2. b. The practice is attested Ø by a number of surviving manuscripts and miniature paintings, ...

(COCA)

The observed affinity of the passive for the zero variant rather than the more explicit variant has been explained with reference to the dispreference for stranded prepositions, which exert a particularly high processing load (cf. Hawkins 1999:276-77, Rohdenburg 2013). This avoidance of stranded prepositions seems to be stronger in passives than in actives.

Minor infelicities concerning copyediting are spelling errors (v, 5, 191), an erroneous reference to examples (18), a punctuation error (40), word omissions (154, 167), missing year of publication (155 n.), and hardly distinguishable shades of coloring in diagrams (181).

In sum, this book is well argued and stimulating. It is highly readable due to its lucid and informative style. Providing indispensable content for those interested in the English passive, it raises new issues concerning motivations for using passives and sheds new light on the representation of agenthood in relation to communicative purpose.