Subordination in English: Synchronic and Diachronic Perspectives 9783110581034, 9783110583571, 9783110581058, 2018940117

Citation preview

Elena Seoane, Carlos Acuña-Fariña, Ignacio Palacios-Martínez (Eds.) Subordination in English

Topics in English Linguistics

Editors Elizabeth Closs Traugott Bernd Kortmann

Volume 101

Subordination in English Synchronic and Diachronic Perspectives Edited by Elena Seoane Carlos Acuña-Fariña Ignacio Palacios-Martínez

ISBN 978-3-11-058103-4 e-ISBN (PDF) 978-3-11-058357-1 e-ISBN (EPUB) 978-3-11-058105-8 ISSN 1434-3452 Library of Congress Control Number: 2018940117 Bibliographic information published by the Deutsche Nationalbibliothek The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data are available on the Internet at http://dnb.dnb.de. © 2018 Walter de Gruyter GmbH, Berlin/Boston Typesetting: Integra Software Services Pvt. Ltd. Printing and binding: CPI books GmbH, Leck Cover image: Brian Stablyk/Photographer’s Choice RF/Getty Images www.degruyter.com

Preface This book deals with subordination in English from both a synchronic and a diachronic perspective. It includes not only studies on British and American English but also on a wide range of EFL, L2 and other learner varieties (with special emphasis on Asian ones), as well as English-based pidgins and creoles. Such a breadth of material should be a source of inspiration for scholars with an interest in historical English syntax, especially those looking at ongoing changes to subordination patterns in a vast range of varieties of English worldwide. Its authors demonstrate the value of theoretically-informed corpus-based approaches as a means of quantifying and describing linguistic change in illuminating and novel ways. Most of the contributions here are based on large electronic databases and corpora of written and spoken English, and the book itself has been made possible by the collaboration of some of the most distinguished specialists in the field, together with a number of up-and-coming scholars. Our sincere thanks to Elizabeth Traugott, the series editor in charge of this volume, for her commitment to the project, her invaluable comments, and her continuous help throughout the process. It was an immense pleasure to work with her. It was also extremely gratifying to work with the fifteen authors who contributed chapters and who helped us with the reviewing process. Their kindness, goodwill and respect for deadlines was admirable. The following colleagues also acted as anonymous external peer reviewers and provided generous feedback, for which we are immensely grateful: Laura Bailey, Laurel Brinton, Sarah Buschfeld, Peter Collins, Clarence Green, Ulrike Gut, Willen Hollmann, Bernd Kortmann, Svenja Kranich, Peter Petré, Paul Rickman, Peter Siemund, An Van Linden and Uwe Vosberg. Our gratitude is also due to the editorial staff of De Gruyter, especially Julie Miess. Thank you all for making this project such an enjoyable and positive one. For generous financial support, we would like to express our gratitude to the Spanish Ministry of Economy and Competitiveness (grants FFI2017-82162-P, PSI2015-65116-P and FFI-2015-64057) and the Regional Government of Galicia (grants ED431C 2017/50, ED431D-2017/09 and GRC2015/006). As editors, this book marks a very special occasion for us. We share a long and rewarding history of scholarly cooperation, all of which has taken place under the expert influence, watchful guidance and meticulous advice of our former doctoral supervisor, Teresa Fanego. Many, many years after we first worked with her, Teresa is about to retire from full-time departmental duties. For all of those who have known and worked with her, the notion of her retiring is simply unimaginable. De facto, Teresa will continue business as usual, which means that we will https://doi.org/10.1515/9783110583571-201

VI 

 Preface

be able to continue to count on her penetrating insights and life-long experience of whatever this academic world that we love may bring our way; and it is a love, we might add, that her example has served to instill in us all. This professional shift, new modus operandi… however we term it, certainly calls for a celebration. If business must go on, and indeed it must, the current volume should reflect the very best qualities of its dedicatee, and thus provide its readership with absorbing, insightful research and just the right degree of theoretical density. We believe that the papers collected herein amply achieve these aims, and we can think of no better way to honor an academic career such as Teresa’s, so full of rigor, precision and obstinate determination never to settle for second best. In sum, we seek to celebrate her sheer love for, and devotion to, the profession, and to express our deep and enduring debt to her for the example which she has set us all. Santiago de Compostela and Vigo, March 2018 The Editors

Contents Carlos Acuña-Fariña, Ignacio Palacios-Martínez and Elena Seoane Subordination, or the permanent allure of the “adjacent possible” 

 1

Part I: Diachronic perspectives on subordination Hendrik De Smet From flying Sancho to swooning Altisidora: The changing use of premodifying present participles in three English translations of Cervantes’ Don Quijote   25 Alexander Bergs Because science! Notes on a variable conjunction  David Denison That-clauses as complements of verbs or nouns 

 43

 61

Juhani Rudanko Semantic roles as a factor affecting complement choice: a case study with data from COHA  85 Günter Rohdenburg Expanding the type you can’t help laughing

 103

Bas Aarts, Sean Wallis and Jill Bowie -Ing clauses in spoken English: structure, usage and recent change 

 129

Hubert Cuyckens and Frauke D’hoedt Resourceful ways of recruiting members: The origin and development of Mental zero-Secondary Predicate Constructions   155 Christian Mair The rise of long catenative constructions in Modern English: new sub-schemas and new stylistic options   185

Part II: Subordination in Present-day Englishes Elisabeth Bruckmaier Catenative get in World Englishes 

 211

VIII 

 Contents

Sven Leuckert Adverbial subordination across variety types: A synchronic analysis of the syntax and semantics of since- and while-clauses in ENL, ESL, and EFL   235 Rahel Oppliger Whatever the specific circumstances, …: A Construction Grammar perspective of wh-ever clauses in English   263 Cristina Suárez-Gómez A sociolinguistic study of relativizers in spoken Philippines English  Subject index 

 309

 285

Carlos Acuña-Fariña, Ignacio Palacios-Martínez and Elena Seoane

Subordination, or the permanent allure of the “adjacent possible” Subordination, along with coordination, is one of the two principal means of combining clauses: unlike coordination, it is the grammatical strategy used to link two clauses asymmetrically. Formally, this asymmetry may reveal itself at the sentence level as a lack of independence of the subordinated constituent, for instance, and may even be heralded by specific markers such as that or to. Functionally, the asymmetry may be reflected at the level of discourse in the guise of information presented as backgrounded or presupposed, that is, not in the current focus of attention (Winter 1982; Langacker 1991: 436; Cristofaro 2003: 30–33; see also Verstraete 2007: ch. 2). As regards how subordination comes into being historically, paratactic structures are typically the main source (Heine & Kuteva 2007; Givón 2009), but English developed an alternative route here: namely, the syntactic reorganization of phrases into clauses that began to occur around 1300, when nominal gerunds started to acquire verbal properties like the ability to code their own direct objects (Fanego 2004a, 2016). From the point of view of the types of subordinated clausal structures we can find, the English system boasts perhaps only six major types: a) that/zero declaratives (see Denison, this volume) and (open and closed) interrogatives (I don’t know whether / when she’ll come, see Huddleston & Pullum 2002: 62); b) relatives (see Suárez-Gomez, this volume); c) bare infinitives (see the chapters by Bruckmaier and Rohdenburg); d) to-infinitives with and without a subject (as in Mair, Rudanko and Bruckmaier, this volume); e)-ings with and without a subject (see De Smet, Mair, Bruckmaier, Rohdenburg, and Aarts / Miller & Bowie); and f) comparatives (It is heavier than I expected, see Huddleston & Pullum 2002: 62). Like most other clause constituents, subordinated clauses may be licenced by a predicate head (I know that she is clever, I want to dance) or occur as some kind of adjunct or adverbial (When walking there, I saw the light; Walking that way, I manage to avoid the pain in my feet; see the articles by Bergs, Leuckert and Oppliger, this volume). The above impressionistic observations somehow obscure the complexity in this area of the grammar of English, which is in fact quite remarkable. First,

Carlos Acuña-Fariña, Ignacio Palacios-Martínez, (University of Santiago de Compostela) and Elena Seoane (University of Vigo) https://doi.org/10.1515/9783110583571-001

2 

 Carlos Acuña-Fariña, Ignacio Palacios-Martínez and Elena Seoane

as is often the case in grammar, there is a continuum that links subordinated and coordinated structures in a fluid way, with catenatives somewhere in the middle but still arguably more subordinate than coordinate (see the chapters by Mair and Bruckmaier, this volume; on the continuum of “integration” see also Lehmann 1988: 183; Hopper and Traugott 2003: 178–179; and Verstraete 2007: ch. 4; among many others). This, however, is not the greatest source of complexity. Perhaps the most interesting property of the system of clausal subordination is that, like the true core of grammar itself, it never stays put. In this volume various authors report on the permanent state of flux that can be appreciated if one examines the historical record (see chapters 2 to 9). By “permanent” we mean that the flux actually goes on as we speak. Thus, for instance, it turns out that from the late Middle English period catenative constructions have increased considerably, both in frequency and in structural diversity, and as a result have become a key part of the Present-day English system of clausal subordination (as reported in Mair, this volume). The gerundial complement has also been spreading, at the expense of the to infinitive. This change is a major aspect of what has come to be called the Great Complement Shift (see Fanego 2004a, b, 2016; Rohdenburg 2006; Vosberg 2006). An interesting consequence of this latter change is another one: the rise of VP deletion marked by to (which, cornered by gerunds, therefore came to specialize in this information compressing strategy; see Rudanko, this volume). Small clauses, as in I consider him a bore, also seem to have been continuously on the rise (see Cuyckens & D’hoedt, this volume), and so have situation-oriented present participles like sobbing boy in He hauled the sobbing boy to his feet (De Smedt, this volume). These participles are premodifying constituents that function as compressed subordinate clauses, relating one (matrix) event to another (subordinate) event which is construed as backgrounded. Their new clause-like behavior is a departure from the adjectival character of premodifying participles, and may be seen as an instance of functional verbalization. As noted, changes never seem to stop, even today: in this volume Denison reports on the recent development from [shell + that] to [V + that], as in X expressed that Middle English was indeed a creole, instead of X expressed the fact that Middle English was indeed a creole, and Bergs documents current ongoing change to the complement structure of the venerable word because, which is currently used in expressions like motivation alone does not assure success because circumstances and Finish this one, but only because famished, with NPs and Adjective Phrases as complements of the preposition respectively. As can be seen, most of these changes involve expansions of the use of pre-existing material which in turn involve different forms of reanalysis. As important as the changes themselves (contemplated as finalized states) are the states of actualization of the reanalysis

Subordination, or the permanent allure of the “adjacent possible” 

 3

(Timberlake 1977, Harris & Campbell 1995: 77, Harris 2003: 532, 536–537), that is, “the gradual manifestation of the innovated underlying structure at the observable level of language use” and the co-habitations that both change and reanalysis enforce for extended periods of time (Fanego 2004b). This can be seen for instance in the constant leakage between gerunds and present participles, a blurred state of affairs causing some scholars to argue that the two clause types have in fact completely merged (Aarts, Wallis & Bowie, this volume), whereas others argue that they continue to be different (De Smet 2014: 229). In such a dynamic system, one can expect rises in use to be accompanied by losses, and indeed these can be easily documented. For example, the spread of the gerundial complement, as noted, took place at the expense of the to infinitive, which became less used. The same spread caused a drastic reduction in the use of genitive subjects (see Rohdenburg and Aarts, Wallis & Bowie, this volume), as the new gerundial forms were being reanalyzed as clauses (not as NPs) and clauses do not usually express their subjects in the genitive. To linguists, documenting all these changes is less important than understanding how they happen. This is because understanding how the system behaves is essential in comprehending the true nature of language. The present volume aims to shed light on the dynamic essence of the system here by discussing a large number of explanatory principles at work. Many of these are wellknown from the grammaticalization and the Construction Grammar literatures. These include the notions of attractor or attractor sets (Traugott 2008: 33), understood as the analogical structure that creates what Kauffman (2003) and Johnson (2010) call “the adjacent possible”. Johnson (2010) describes this as follows: “The adjacent possible is a kind of shadow future, hovering on the edges of the present state of things, a map of all the ways in which the present can reinvent itself”. As Cuyckens & D’hoedt note (this volume), for language change, the “adjacent possible” means that at any given moment in time only a small number of changes are possible, each one potentially leading to new opportunities for language to be reinvented in a motivated kind of way. In a framework of this kind, the nature of multi-sourcing (i.e. where multiple constructions interact to create a new construction, see Van de Velde et al. 2013), needs to be accounted for. Multi-sourcing naturally leads to the notion of inheritance (Goldberg 1995, among many others) and this in turn to network building, all of these being indispensable formants in the construction of grammar, of which the grammar of clausal subordination in particular makes ample use. For instance, Traugott & Trousdale (2013) have convincingly argued that the present-day way-construction in English (e.g. Tom elbowed his way to the top in politics) is the result of two different constructions feeding into it. The first involved intransitive verbs of motion that combined with way as an adverbial. The second was the use of way as the object of a transitive

4 

 Carlos Acuña-Fariña, Ignacio Palacios-Martínez and Elena Seoane

verb with the basic meaning of creation or acquisition of a path. Multi-sourcing, inheritance and attractor sets make these historical moves possible. Interestingly, multi-sourcing is both very difficult to track and tempting to invoke. Yet, very often a distinction must be made between two or more constructions that contribute features to another construction and the pre-existence of a construction which simply acquires new members. Thus, for instance, Cuyckens & D’hoedt (this volume) observe that small clauses in English did not actually expand in the way that they did because they recruited features from neighbouring constructions or merged with them in novel ways. Instead, they simply accepted new verbs of the Mental type of predicates (think, consider, etc.), thus augmenting both their type and token frequencies. They term this mechanism “categorial incursion” (De Smet 2009; Petré 2012, 2014). The authors point out that, in the case of small clauses, categorial incursion worked in recognizable steps: first it was random and unsystematic polysemization in OE; this was followed by increasing systematicity in ME; and finally Romance influence occurred throughout ME and EModE. This scenario brings to bear another set of explanatory principles. There is in the first place a process of polysemization which allows verb types in the Small Clause Construction to become polysemous, acquiring mental meanings. There is also metaphorization in the process, whereby the (semi-)external meaning of a predicate (like see) leads to the expression of internal meaning (“consider”/”think”). Polysemization and metaphorization are made possible by “analogical snowballing” (Ogura & Wang 1996) in the sense that strong attractor sets in the mental class exerted an influence on new members through analogy. Analogical snowballing is an incremental, self-feeding process that has the stamp of many cognitive operations: it is fed by frequency, and it produces ripple effects. Both multi-sourcing and categorial incursion are constructionalization principles to be distinguished from seemingly similar principles like opportunistic exaptation (Lass 1990, 1997). As Lass (1997: 316) notes, exaptation “is a kind of conceptual renovation, as it were, of material that is already there, but either serving some other purpose, or serving no purpose at all. Thus perfectly ‘good’ structures can be exapted, as can junk of various kinds”. For instance, Croft (2000: 129) has convincingly argued that in some nonstandard varieties of English the (largely useless) 3rd-person -s suffix has already been exapted for use with verbs coding specific or single events (e.g. I sees the architect tomorrow), and thus contrasts with other verb forms expressing iterative or habitual actions (e.g. I do see her every day). In the same vein, and more to the point here, Fanego (2004b) has argued that the pattern the deceiving him was easy, which arose in Late Modern English, shows an exapted use of the article the as a complementizer. Interestingly, Fanego also studies an ongoing change, that of “impertinent by”, as in by trying to make her mother happy proved unlucky for Paul, which is apparently

Subordination, or the permanent allure of the “adjacent possible” 

 5

becoming frequent among American undergraduate students in contemporary English (Hogan & Hogan 1998). She argues that the presence of both the and by in their respective structures and time frames is motivated by the same principle of transparency: the need to provide explicit grammatical marking for clauses that function as (complex) pre-verbal subjects. Unlike extinct, exapted, complementizer the, however, “impertinent by” did arise via a grammaticalization route involving classic processes of desemanticization, extension or context generalization, and decategorialization. Attractors in the confines of the adjacent possible, multi-sourcing, inheritance, categorial incursion, polysemization, metaphorization and exaptation apply in principle to both complement and adverbial clauses that are asymmetrically linked to a matrix state of affairs, even if these are loosely integrated into the structure of the composite state of affairs (see Verstraete 2007: 172 & 227, and references therein). Take wh-ever clauses, for instance, as discussed by Oppliger in this volume. These detached clauses show symptoms of constructionalization of the kind also discussed by Traugott & Trousdale (2013). The first of these happens when the interrogative pronouns and the temporal adverb ever fuse in spelling and develop concessive connectivity. Second, the peripheral position of wh-ever clauses expands both their syntactic and their pragmatic scope: “the wh-ever clause operates on the periphery, rather than as a constituent of the sentence, its scope extending beyond the sentence, to the discourse level” (Oppliger, this volume). In fact, König and van der Auwera (1988: 128) call these constructions “speech act qualifying adverbial clauses” which are “about rather than a part of the main clause”. Their peripherality allows them to code speaker attitude or viewpoint (subjectivity), particularly the I take this as beside the point reading that characterizes them so strongly. So the wh-ever clause construction involves both a reduction – of the form of the connective element – and an expansion – of its syntactic and especially pragmatic scope. Additionally, wh-ever clauses are schematic in nature (both the form of the verb in them and the Wh-word itself can vary), containing more schematic meso-constructions with variable slots and more specific micro-constructions with fixed lexical elements. And they fit into a radial structure containing larger groups of subordinate clauses, with which they share certain meaning components. Thus, for instance, because they share characteristics with both concessives and conditionals (they are in fact concessive conditional clauses which present an antecedent condition that is irrelevant to the truth conditions of the consequent main clause), it is understandable that they are often treated as a subgroup of either class of clauses (König 1985: 3; see also Haiman 1974 and König & van der Auwera 1988: 107). A cautionary warning is implicit in all the previous considerations (see Fanego 2004c too). In view of the complexity of the system that generates

6 

 Carlos Acuña-Fariña, Ignacio Palacios-Martínez and Elena Seoane

the grammar of clausal subordination in English, it stands to reason that one should view with suspicion any attempt to set the largest part of the explanatory mechanisms involved in a purely conceptual structure. Take sentential complements, for instance. Most work in this area has long recognized that matrix verbs with similar or related meanings (such as remember and forget) will typically take the same kinds of complements; that verbs denoting volition (wish, want, etc.) usually combine with to-infinitives; that certain -ing clauses tend to be associated with factuality, and to-infinitives with non-factuality; and that non-deranked clauses (e.g. that-clauses) typically collocate with cognition predicates such as believe or communication predicates such as point out (Quirk et al. 1985: 1169–1221; Biber et al. 1999: 753–759; Huddleston and Pullum 2002: ch. 14; Miller 2002: 44, 47; Verstraete 2007: ch. 11; among many others). Langacker (1991: 35) has taken the conceptual underpinnings of a semantically-motivated system of this kind to its extreme by suggesting that, in the end, the choice of that is “to step back from the situation […] and construe it as an abstract object or proposition capable of being manipulated, evaluated, and commented on. Instead of being asserted, this proposition is taken as one participant in a higher order relationship […], whence its role as a clausal subject or object”. Construing the subordinate event as a main-clause participant implies a conceptual distancing whereby this event is captured holistically through summary scanning (not sequential scanning, crucially) and manipulated as a unitary percept (Langacker 1991: 439; 1992: 305–306). For this reason, Langacker argues, complementizers (i.e. -ings, to, or that) “are plausibly analyzed as imposing an atemporal, perhaps even a nominal construal on the structures they combine with” (1991: 440). As Fanego (2004c) has suggested (see also Miller 2002: 345 among others), this elegant line of reasoning – and similar ones – leaves the entire Complement Shift unaccounted for. We might consider, for example, the development of the nominal gerund into a verbal form from about 1300 onwards. It makes sense to understand that this change was either initiated or encouraged by systemic pressure to develop a clausal pattern capable of occurring after prepositions. Over the centuries, snowballing caused verbal gerunds to appear in non-prepositional contexts and to become common as subjects, objects, and predicatives, yet as Fanego (2004c) notes, even today gerunds dependent on a preposition represent around 68 per cent of all gerundial uses. No obvious conceptual substrate is visible here. What is visible is a dynamical system that self-structures, beginning with what material it has at hand, and develops opportunistically using the cognitive operations referred to above: analogical snowballing, multiple inheritance, metaphorization, exaptation, etc. There is in fact a strand of work within the dynamical systems methodology (Van Orden et al., 2005; Kloos & Van Orden,

Subordination, or the permanent allure of the “adjacent possible” 

 7

2009; Rączaszek-Leonardi, 2012, 2014) that emphasizes dependence on initial states, interconnectedness of subsystems, the emergence of attractor states, and the applicability of complex systems methods in order to understand change over time. Given that the development of at least some dynamical systems is highly dependent on their initial state, seemingly insignificant differences at the beginning of a process may have substantial consequences in the future development of the systems (the so-called Butterfly Effect). Such dynamical systems of co-adapted traits are therefore in permanent flux. This puts not only language evolution, but probably even learning and use, in the domain of synergies and self-organization (Kloos & Van Orden, 2009). In fact, processes which are often seen as of a modular type may instead be seen as part of a complex, unified, dialogical universe (Acuña-Fariña 2016). It seems that a fair conclusion one may draw from the range of phenomena analyzed in this volume is that the grammar of clausal subordination fits into a view of language as a dynamic object of this kind. The previous observations seem to imply a reification perspective on language (language is responsible for change), as opposed to a usage-based one (speakers are responsible for change). Indeed, this is very conspicuous in statements of the kind “the dynamical system self-structures”, but it is even more implicit in attenuated expressions like “verbs become polysemous and acquire meanings” (see above). On other occasions, however, the reader may have noticed that we have adopted a more usage-based approach here, as in our brief description of exaptation, where we made the point that the 3rd-person -s suffix in English has been “exapted for use” with certain verbs (see above). Most of the content of this book is framed in a way that is actually more compatible with the view that “systemic pressures” are best seen as a tendency on the speakers’ part to make large generalizations and analogize on the basis of them, that is, with a usagebased perspective. It is fair to add, though, that neither the whole book itself nor this introduction aim to focus on the reification vs. usage-based debate, at least not directly. The contributions that follow provide an exhaustive account of the range and depth of possibilities that can be observed in the vast domain of clausal subordination in English. Section I contains diachronic approaches to different subordinated constructions, while Section II deals with subordination in Present-day Englishes. The opening chapter discusses compressed subordinate clauses, in particular with situation-oriented premodifying participles. In “From flying Sancho to swooning Altisidora: The changing use of premodifying present participles in three English translations of Cervantes’ Don Quijote”, Hendrik De Smet shows the increase in the number of situation-oriented present participles (e.g. stretcht upon the burning sand) over the Modern English period, which represents,

8 

 Carlos Acuña-Fariña, Ignacio Palacios-Martínez and Elena Seoane

he argues, a recent case of functional verbalization. To come to this conclusion, and as a follow-up study of previous research on different material (De Smet and Vancayzeele, 2014), De Smet analyzes untranslated fiction, a corpus of over 1.6 million words formed by the original Spanish masterpiece and three successive translations dating from 1612–1620, 1885 and 2003 respectively, corresponding to roughly Early Modern, Late Modern, and Present-day English. The analysis is carried out both quantitatively and qualitatively by presenting first general frequencies in the use of these present particles in the different translations/periods and then by studying in closer detail the function of each token selected from the original and from the corresponding translation. It should also be borne in mind that in Spanish the proportion of premodifying present participles is not very high and, as a result, those that occur in the English data may correspond to a combination of Spanish source patterns and, most probably, to a more idiomatic, rather than a more literal, translation of the original. This means that the occurrence and use of these participles are not conditioned by the source language but by the resources employed by the translators. As mentioned, the results not only show that premodifying present participles increase in number in the different English translations of Cervantes’ Don Quijote, but also reveal that this tendency applies to the group of situationoriented participles. With respect to the previous study, the author highlights that the frequency of present participles is considerably lower. Crucially, the data reveal that there is a notable change in the functions of these situation-oriented participles, which generally construe a process as being backgrounded but simultaneous to the more salient event of the superordinate clause. It is also observed that the different translations increasingly use situation-oriented participles in contexts where there is not always a correspondence with the original, that is, not in all cases is there a simultaneity in the actions of the participle and the main clause. In the first translation these participles are scarcely used and when they are, there is a reason for it. In contrast, this is not the case in the two later translations, where they tend to be more freely employed and without being necessarily faithful to the original. As a conclusion, De Smet maintains that situation-oriented participles have become a useful linguistic resource for authors and translators, and this change could be easily considered as a case of functional verbalization. Also from a diachronic perspective, Alexander Berg’s contribution, “Because science! Notes on a variable conjunction” provides a Construction Grammar approach to a number of innovative uses of the subordinating conjunction because in spoken, written and electronic media. The “new” because is followed by different complements, namely NPs, AdjPs, interjections, pronouns, etc., and these constructions are compared and contrasted with more general or

Subordination, or the permanent allure of the “adjacent possible” 

 9

regular structures in which because precedes of or a full clause. Bergs thus compares structures such as he went home because of a headache vs. he went home because headache. His data, extracted mainly from the Corpus of Historical American English (COHA) and the Corpus of Contemporary American English (COCA), show that, contrary to what might be expected, these particular constructions of because cannot be regarded as totally new or recent, since in the nineteenth century tokens of the because +AdjP construction are already recorded, and examples of the because + NP combination, although relatively rare, can also be found at least from the 1940s onwards. According to Bergs, because X and causal because are semantically similar constructions although pragmatically different. Style and register seem to play an important role in this respect, since because + X is much more common in spoken, on-line and informal contexts than causal because. Furthermore, communicatively they express different functions. There is a difference in subjectivity since the because X construction entails a higher degree of speaker subjectivity than the corresponding causal because construction. Bergs argues that from the point of view of Construction Grammar because X is a meso-construction that interacts with at least three microconstructions: because N, because A and because particle. He then assumes that there was a gradual development and spread of this construction, emerging with a few particular tokens of the because + type which led to the formation of the more general and micro-construction because + A, which in turn favored the creation of new because + A constructs. As an extension, and in analogy, because + N was formed following a similar process and evolution to the previous one, and the same applies to because + particle. The paper concludes with a number of reflections on the need to re-evaluate the concept of subordination, especially in spoken language, where it is common to find incomplete structures which cannot always be classified as typically subordinate. An area of subordination that finds favor in previous research is that of clausal complementation. In this volume, six chapters testify to the continued interest in the clausal complementation of nouns, adjectives and verbs among English historical linguists and, more recently, among scholars working on World Englishes. These are the contributions by Denison, Rudanko, Rohdenburg, Mair, Aarts, Bowie & Wallis, Cuykens & D’hoedt (Section I) and Bruckmaier (Section II). In “That-clauses as complements of verbs or nouns”, David Denison deals with the recent development of “shell nouns” + that to V + that structures, that is, from abstract (shell) nouns such as fact, claim, argument and situation, followed by that structures (cf. it highlighted the fact that X) to the new, post-2000 corpora V + that structures (cf. it highlighted that X). This development involves the extension of that-complement clauses to verbs that had never shown this pattern, such

10 

 Carlos Acuña-Fariña, Ignacio Palacios-Martínez and Elena Seoane

as advance, contest, contradict, endorse, highlight, moot, pose, put forward, propound and uncover. For the purpose of this analysis, Denison extracts material from COHA, which is contrasted and complemented with information taken from general reference grammars, dictionaries and additional searches of the OED. For practical reasons, the search is limited to the twentieth century plus the first decade of the twenty-first. The analysis of the data shows that by 1900 the construction V + that was largely consolidated in the language, although different sets of words tended to behave differently over time. Thus, a total of 24 verbs such as admit, assert, consider, declare, demand and explain are already recorded at the beginning of the twentieth century and their frequency remains high after the 1920s. In addition, a group of seven other verbs (accept, concede, emphasise, establish, repeat and stress) were not fully entrenched in 1900 although they became quite frequent later on. It is also observed that a group of verbs from the list selected which did not obtain many hits in the 1900s or 1910s did not become more frequent later either. Finally, a set of verbs including appreciate, clarify, document, hate, highlight, like and love showed a notable rise over time. When comparing the verbs + that construction with the shell noun + that, a moderate decline is detected, although some colloquial verbs, such as document and posit, show a recent growth in frequency. In light of this data, Denison concludes that the two constructions are semantically similar although the V+ that is regarded as more economical, which may account for its wide distribution. Denison provides two possible explanations for this syntactic evolution: one is the mechanism of analogy and the other is the lack of rigor in the use of online thesauri, which tend to provide synonyms of words without any information on complementation patterns. In another chapter on complementation, “Semantic roles as a factor affecting complement choice: A case study with COHA”, Juhani Rudanko explores the rivalry between to infinitive and to -ing as complements of the adjective unaccustomed in the nineteenth and twentieth centuries, whereby to -ing complements (she is unaccustomed to eating early) surpass to infinitive ones (she is unaccustomed to eat early) in the twentieth century. He finds the variation between the two subordinate complementation patterns and their change in prominence to be an instantiation of the Great Complement Shift, amply studied by Fanego (1996, 2016) among others. Data from COHA show that the critical period of variation is 1850–1930, and Rudanko focuses on this period to inquire into the factors that may have had a role in this complement shift. He reports that, for example, the Extraction Principle (Vosberg 2003) has a limited role in explaining the variation found, whereas the Choice Principle (Rudanko 2017) does seem to shed light on complement selection in the period under analysis. More crucial for the relative decrease of to infinitives after unaccustomed, it is argued, is the change in the

Subordination, or the permanent allure of the “adjacent possible” 

 11

nature of infinitival to, which acquires auxiliary status in the nineteenth century, as evidenced by the dramatic spread of VP Deletion constructions with to infinitives (as in he is eager to go but his sister is not eager to). Rudanko argues that the more nominal character of to -ing complements, where to is a preposition, could have facilitated the increase in frequency of to -ing after unaccustomed, since this adjective can also take to NP complements (she is unaccustomed to fights). Like Rudanko, Günter Rohdenburg examines diachronic variation and change in subordinate non-finite complementation patterns. In “Expanding the type you can’t help laughing”, he explores a network of low-frequency expansions of the subordinate negative polarity constructions help -ing (as in the title of the paper) and help but, as in you can’t help but laugh. The study is based on the British National Corpus (BNC), COHA, and a 3,151 million-word collection of British and American newspapers covering the 1990s and early 2000s. The corpora show that the emergence of basic help -ing and help but predate their expansions, both in British and American English, and that simple expansions precede the development of more complex ones. The expanded categories involve, first, the inclusion of various types of nominal subjects, such as reflexives (he can’t help himself talking / we can’t help ourselves but talk), which, in the case of help -ing, have gained prominent status among the gerunds involving an overt subject. Other types of nominal subjects are personal pronouns and full NPs. The latter two categories are only possible with help -ing, and both the pronoun and full NP subject may take possessive or genitival form (I couldn’t help his / the child’s being here), but these forms decline from Late Modern English in favor of object case pronouns (him) and non-genitival NPs (the child, see also Aarts, Wallis & Bowie, this volume). Secondly, another expansion is the establishment of the optional from-gerund, dating from the nineteenth century, which is typical of a large number of other negative implicative verbs such as discourage and refrain (as in she could not help the protest from leaving her lips). Rohdenburg shows that the distribution of transitive and intransitive uses of help -ing with respect to the incidence of from falls in line with the other negative implicative verbs and also follows the prediction of the Complexity Principle, in that transitive uses of help -ing, which are more complex than intransitive ones, clearly favor more explicit gerunds using from. He also shows that the well-known preference in American English for gerunds introduced by from does not become visible until the twentieth century. Thirdly, the last expansion examined is the (rare) use of be able to supplementing can or could where these cannot occur, such as in combination with another modal verb (cf. Drivers may not be able to help speeding). In the conclusions, Rohdenburg can’t help but wonder why other apparently isolated developments have occurred, such as the rare use of reflexive help but in contrast to the increasingly common reflexive help -ing.

12 

 Carlos Acuña-Fariña, Ignacio Palacios-Martínez and Elena Seoane

Subordinate -ing clauses are taken up again in the next chapter, “-Ing clauses in spoken English: Structure, usage and recent change”, by Bas Aarts, Sean Wallis & Jill Bowie. They explore -ing verbs heading subordinate clausal structures in spoken Present-day English (e.g. The navy deliberately sinking the ship, where sinking takes a subject and an object), which are different from -ing verbal nouns heading NPs (as in The deliberate sinking of the ship). Aarts, Wallis & Bowie provide a comprehensive account of the recent development of -ing clauses based on the Diachronic Corpus of Present-Day English (DCPSE) , which shows a number of changes in this recent and relatively short period of time (1955–1995). On the one hand, by plotting the share held by each function that an -ing clause can perform (the proportion p(function | -ing)) they find a significant increase in the proportion of -ing clauses functioning as subjects (especially without a subject) and direct objects (especially with a subject), as well as a diversification of functions over time. On the other hand, they demonstrate that -ing clauses have become more clausal over time in spoken British English, since, taken together, they increasingly tend to contain a subject. This is especially the case with those taking the function of direct object and adverbial of a matrix clause, as in I hope you don’t mind my rubbing my hands. The implication of this finding is that the “verbalization” that Fanego (1998), among others, have documented for Middle English would still be at work in Present-day English. More evidence suggesting an increasing “clausalization”, as they call it, of -ing clauses is the fact that the frequency of genitival subjects – an indicator of nominal status – is decreasing. Aarts, Wallis & Bowie agree with De Smet (2010) that there is a historical trend for gerunds and -ing participles to become less distinctive, and their data lead them to suggest that the merger process between both is heading towards completion. Consequently, they question whether the label “gerund” is necessary for the description of the grammar of Present-day English. Hubert Cuyckens & Frauke D’hoedt discuss the development of [Verb + NP + XP] sequences such as I find him very unrealistic, that is, Mental zero-Secondary Predicate Constructions (Mental zero-SPCs) or “small clauses” (see for example Los 2005). The article’s title, “Resourceful ways of recruiting members: The origin and development of Mental zero-Secondary Predicate Constructions”, alludes to the long and winding history of Mental zero-SPCs, which they carefully trace in this paper. In the suite of the Penn Corpora of Historical English and the YorkToronto-Helsinki Parsed Corpus of Old English Prose they find that from Old to Late Modern English Mental zero-SPCs have become more frequent in token frequency and also more productive in terms of type frequency. Cuyckens & D’hoedt argue that the construction comprises many different types of verbs as the result of a process of internalization (cf. Traugott 1989) whereby external verbs (causative/ labelling verbs such as render and call) gradually gave way to more internal ones

Subordination, or the permanent allure of the “adjacent possible” 

 13

(mental/cognition/volition verbs such as think, deem and wish). Cuyckens & D’hoedt set out to find where the new members have come from, and what underlying processes account for their recruitment into the mental representation class. They report that in OE new members were recruited though random and unsystematic polysemization, with new members originating from different verb classes (perception, causative, communicative and possession verbs) and undergoing a process of metaphorization and internalization facilitated by the presence of a reflexive pronoun; no clear “attractor sets” or analogical models for other non-mental verbs are detected in this period. This process became more systematic in ME, with the secondary mental verbs hold and find acting as an attractor set after which other non-mental verbs are analogically modelled. Together with attraction of core mental verbs, in ME another mechanism of membership recruitment is identified, which extended well into Early Modern English, namely attraction (and mass addition of) verb types of Romance origin. In Early Modern English, both type and token frequency reach stable values, since the mental verb class exhausted most mental verbs of the language. Their data thus indicate that the construction has expanded essentially through the mechanism of “categorial incursion”, that is, the introduction of new verbs into a category and construction that already exists. Therefore, the complex interplay of mechanisms shaping the development of Mental zero-SPCs do not establish a construction that is entirely new to the language, and for this reason it cannot be said to constitute a multi-source construction in the strict sense. Among complementation patterns, two-element catenative constructions (e.g. this helps (us) (to) pay the rent) are a cornerstone of the present-day English system of clausal subordination given their considerable increase in frequency and structural diversity from the late Middle English period. This is what Christian Mair documents in his contribution “The rise of long catenative constructions in Modern English: New sub-schemas and new stylistic options”, in which he focusses on the recent history of three- and four- element catenative sequences, as in (…) help the city avoid having to sell additional securities. In his paper, he charts the increase of three-element catenative sequences for the nineteenth and twentieth centuries on the basis of data from COHA, using the example of make NP want to VERB as illustrative of how this sequence moves away from the compositional semantics of the early occurrences “towards constituting a separate low-level verb-specific constructional sub-schema in its own right.” The constructionalization process also manifests itself in new collocational profiles identified in the twentieth century, especially those in which the verb following want expresses semi-conscious physical urges and violent activities which are not actually carried out (e.g. it made me want to cry / run / puke / kill / vomit). As for four-element catenative chains, he explores them in COHA and

14 

 Carlos Acuña-Fariña, Ignacio Palacios-Martínez and Elena Seoane

COCA and finds a recent increase in frequency and a preference for certain recurrent combinations, especially those built around sequences of causative (make, let, get) and aspectual verbs (start, continue, stop, e.g. let her begin to learn to do). Mair also notes the use of more marginal members of the causative class, not only make and let but also allow, compel and enable. For four-element catenatives, no straightforward constructionalisation can be identified, but they do show signs of systematic conventionalization and seem to have become stylistic usage conventions of Present-day English. Their increasing occurrence, however, cannot be justified by invoking the speakers’ desire to compress information (Biber & Clark 2002), since they are as common in spontaneous speech as they are in writing. The increasing length and complexity of these sequences, which is at times functional and at other times merely ornamental, lead Mair to carry out a contrastive and crosslinguistic comparison. He undertakes this by looking at German and Spanish translations of English long catenatives using InterCorp. In general, the Spanish versions tend to render all the propositional components of the English version, whereas the German versions are reductive and free. Overall, the contrastive study serves to identify functional equivalents of English long catenative chains in these two languages and to bear out the default assumption that constructions and contructionalization are language-specific. Catenatives are also the subject of the next contribution, Elisabeth Bruckmaier’s “Catenative GET in World Englishes”. This is the first chapter in Section II of this volume, which deals with subordination in contemporary English. Bruckmaier broadens the scope here to include different varieties of English, namely British (BrE), Jamaican and Singapore English, as represented in the Lancaster-Oslo/Bergen Corpus (LOB), the Freiburg-LOB Corpus (FLOB), the International Corpus of English (ICE) and the Global Web-based English Corpus (GloWbE). After a discussion of what qualifies as a catenative verb, Bruckmaier analyzes the various forms and meanings that catenative GET in particular can take and their distribution across the three varieties of English. Corpus evidence from LOB, FLOB and ICE shows that catenative GET is more frequent in the New Englishes, Jamaican and Singapore English, than it is in BrE, and that it is also more frequent in written than in spoken language. She draws a basic distinction between simple catenative (GET + verb, as in I got to like her) and causative catenative (GET + NP + verb, as in Get her walking) and reports that simple catenative GET is prevalent in the New Englishes, which might be expected given the tendency towards simplicity typical of New Englishes, while causative catenative GET, which shows more complexity, is predominant in BrE. Her systematization of catenative GET constructions distinguishes three basic meanings, permission, achievement and ingressive meanings, and six formal configurations; an example would be GET to Vdynamic, as in I got to dance with him, which can express permission or achievement.

Subordination, or the permanent allure of the “adjacent possible” 

 15

Exploration of the corpus data shows that, overall, permission and achievement meanings are characteristic of catenative GET in Jamaica and Singapore English whereas the ingressive meaning is typical of the British variety. As for the competition between -ing and to infinitive complementation (I got him to do / I got him doing), -ing complements feature more prominently in BrE than in the New Englishes. Bruckmaier then proceeds to explore GloWbE, and her findings do not corroborate the divide found in ICE between the British variety and the two New Englishes: in GloWbE Jamaican English does not pattern like Singapore English, but is closer to British values in most of the variables analyzed. For example, Jamaican English in GloWbE shows much lower catenative GET frequencies than Singapore English (and also lower than in BrE). As noted, regarding the competition between the -ing form and the to infinitive after GET, the British variety favors the use of -ing more frequently than the New Englishes, and this is the only variable that is confirmed in all corpora. Bruckmaier detects several potential innovative patterns, all of them taking place in Singapore English: GET NP bare infinitive (get them do, probably an analogy with HAVE NP bare infinitive), got and bare infinitive (got go Fantasy Island before, interpreted as resulting from the influence of Colloquial Singapore English) and GET and bare infinitive (you get buy a new car), with a modal meaning similar to that of can. Like Bergs’s chapter on the evolution of because, Sven Leuckert’s contribution is concerned with two adverbial subordinators, in this case since and while, in three different varieties of English: English as a Native Language (ENL), English as a Second Language (ESL), and English as a Foreign Language (EFL). “Adverbial subordination across variety types: A synchronic analysis of the syntax and semantics of since- and while-clauses in ENL, ESL, and EFL” is an innovative undertaking in that it combines a corpus-based and a contrastive approach in the study of the three varieties. For this purpose, ICE-Great Britain is regarded as representative of ENL while the corresponding components of the ICE corpus, ICE-Hong Kong and ICE-Singapore, are selected as prototypes of ESL. The EFL samples were extracted from the International Corpus Network of Asian Learners of English (ICNALE), a database of contrastive interlanguage, which contains spoken and written data produced by Asian learners of English. A total of 194 and 258 clauses introduced by since and while, respectively, were carefully annotated and subsequently analyzed according to the following criteria: variety of English (British English, Hong Kong English, Singapore English, EFL in China, EFL in Indonesia, EFL in Japan, EFL in South Korea, EFL in Thailand and EFL in Taiwan), variety status (ENL, ESL, EFL), meaning (for while, concession, contrast and time; for since, cause and time), position of the clause with respect to the matrix clause (initial, medial, end), medium of expression (spoken vs. written), setting (formal vs. informal), and finally length, defined in terms of the number of words in the subordinate

16 

 Carlos Acuña-Fariña, Ignacio Palacios-Martínez and Elena Seoane

clause following the subordinator. The survey includes both raw and relative figures for these two types of adverbial subordination which provide a qualitative overview of the most innovative features of these two subordinators together with a more quantitative examination. The results obtained suggest that since, on some occasions, replaces other subordinators such as when in learner language, although it is not clear whether cases like these can be regarded as real innovative features or, on the contrary, they should be analyzed as learner errors. It is also curious to see that, as regards adverbial subordination, Asian learners of English and Asian speakers of ESL follow some common patterns, although differences are also recorded in some others. Moreover, ENL and ESL subjects behave similarly for while-clauses; however, it is in the case of since-clauses where most coincidences are found between ENL and EFL. Finally, the position factor in while-clauses seems to be conditioned by meaning while variety status is the defining variable in those introduced by since. Another paper dealing with adverbial subordination is Rahel Oppliger’s “Whatever the specific circumstances, … – A Construction Grammar Perspective of wh-ever clauses in English”. She examines sentence-initial concessive conditional wh-ever clauses – introduced by however, whatever, whenever, whichever and whoever – from a Construction Grammar perspective. These clauses generally present a foregrounded condition or situation that is irrelevant to the truth condition of the consequent main clause. Oppliger is interested in investigating the structural variability, in particular ellipsis, within these clauses, and the organization of constructions within their general system according to their varying schematicity. For this purpose, she uses data from COCA, selecting the academic and magazine samples. More than 2,000 tokens are retrieved and coded for whether a finite verb and an epistemic modal are present in the wh-ever clause or not. The connection between the antecedent and the main clause with a pronominal co-reference is also another structural feature investigated. The results show six different types of wh-ever clauses, with the group of whatever clauses obtaining the lowest presence of verbs in the two registers and with less than half of them having a finite verb in academic prose. This contrasts with the rest of the wh-ever clauses which in most cases occur with a finite verb across the two registers considered. Differences are also reported regarding the presence of an epistemic modal, may or might, in these sets of clauses. While those initiated by whenever, wherever, whichever and whoever very rarely contain an epistemic verb, such a verb is found very frequently in however and whatever clauses. Verb ellipsis is also very common in all wh-ever forms, and adjectives such as possible, feasible, appropriate and necessary are also very frequently found in whenever and wherever clauses. Conversely, verbless whatever clauses occur only

Subordination, or the permanent allure of the “adjacent possible” 

 17

with noun phrases. Furthermore, the figures obtained indicate that register does not have any bearing on the connection between antecedent and main clause in wh-ever clauses. In the last part of the paper wh-ever clauses are examined by applying Traugott’s (2007) constructional hierarchy, and in this respect wh-ever clauses are described as constructions at varying levels of schematicity, that is, as macro-constructions, meso-constructions, micro-constructions, or simply as constructs. Thus some of these wh-ever clauses, such as wh-ever + NP/ADJ + finite verb/epistemic modal/verbal ellipsis or whenever/wherever+ ADJ + verb ellipsis, can be situated on the mesocontruction level while some others, like whatever + the case+ verb ellipsis or whenever + possible + verb ellipsis, function as lexical bundles and can be analyzed as micro-constructions. Finally, Cristina Suárez-Gómez’s study is the only contribution in the volume concerned with subordinate relative clauses, an area of subordination that has attracted much attention both from a diachronic and a synchronic point of view. In her chapter, “A sociolinguistic study of relativizersin spoken Philippines English”, Suárez-Gómez adopts a sociolinguistic perspective to explore the distribution of relativisers in Philippine English (PhiE) as compared to other Asian English varieties, such as Hong Kong English (HKE), Indian English (IndE) and Singaporean English (SgE). For this purpose, she analyzes over 900 examples of adnominal relative clauses introduced not only by explicit relativizers but also by zero. Her results are based on data drawn from the Philippines component of the International Corpus of English (ICE-PHI). Her findings indicate that, in terms of absolute frequencies, the distribution of relativizers in PhiE conversations follows this pattern: who >that >zero >which. This contrasts with what has been found in previous studies (Collins et al. 2014) in written language, where that and which occupied the first positions followed by who, whose and whom. When compared to the other varieties of Asian English, she finds that PhiE resembles IndE in that who is the dominant option in both cases and it shares with HKE and SgE the selection of which as the least frequent of all the four relativizers. As regards the contextual preferences of relativizers, who is the favored alternative when referring to human antecedents while that is reserved for non-human. In terms of the syntactic role of the relative, who is the only relative functioning as subject with human antecedents while in the case of non-human that is the preferred option. In the case of adverbials, adverbials of reason in PhiE are mostly introduced by why, a tendency that is also observed in the other Asian varieties, and the same applies to adverbials of manner where zero is the preferred option. This contrasts with adverbials of place, which are generally introduced by where and a typical Philippine form, wherein. In addition to this, Suárez-Gómez draws our attention to the existence of a number of structures which are not found in standard English: the use of which in combination

18 

 Carlos Acuña-Fariña, Ignacio Palacios-Martínez and Elena Seoane

with animate antecedents, clauses with what introducing adnominal relative clauses, and structures in which a stranded preposition is missing. The paper concludes by acknowledging the existence of a number of idiosyncratic features in PhiE which may support the hypothesis that this new English is developing its own endormative patterns in keeping with what is advocated by Borlongan (2016), who maintains that PhiE has already advanced to phase 4 of “endonormative stabilization” according to Schneider’s (2007) Dynamic Model. With the contributions in this volume, we hope to have conveyed the breadth and depth of the vast domain of clausal subordination in English, both in Present-day and in earlier English, and we also hope that they inspire new avenues for future theoretical and empirical work on English subordination, as much exploratory and explanatory work remains to be done.

References Acuña-Fariña, Carlos. 2016. Opportunistic processing of language. Language Sciences 57. 34–48. Biber, Douglas & Victoria Clark. 2002. Historical shifts in modification patterns with complex noun phrase structures. In Teresa Fanego, María López-Couso & Javier Pérez-Guerra (eds.), English historical morphology: selected papers from 11 ICEHL, Santiago de Compostela, 7–11 September, 2000. Amsterdam: Benjamins. 43–66. Biber, Douglas, Stig Johansson, Geoffrey Leech, Susan Conrad and Edward Finegan. 1999. Longman Grammar of Spoken and Written English. Harlow: Longman. Borlongan, Ariane M. 2016. Relocating Philippine English in Schneider’s dynamic model. Asian Englishes 18 (3). 232–241. Collins, Peter, Xinyue Yao and Arianne M. Borlongan. 2014. Relative clauses in Philippine English: A diachronic perspective. In Lieselotte Vandelanotte, Kristin Davidse, Caroline Gentens and Ditte Kimps (eds.), Recent Advances in Corpus Linguistics: Developing and Exploiting Corpora, 125–146. Amsterdam: Rodopi. Cristofaro, Sonia. 2003. Subordination. Oxford: Oxford University Press. Croft, William. 2000. Explaining language change. An evolutionary approach. Harlow, Essex: Longman/Pearson. De Smet, Hendrik. 2009. Analysing reanalysis. Lingua 119 (11). 1728–1755. De Smet, Hendrik. 2010. English -ing-clauses and their problems: the structure of grammatical categories. Linguistics 48 (6). 1153–1193. De Smet, Hendrik. 2014. Constrained confusion: the gerund/participle distinction in Late Modern English. In Marianne Hundt (ed.), Late Modern English Syntax. Studies in English Language, 224–238. Cambridge: Cambridge University Press. De Smet, Hendry and E. Vancayzeele. 2014. Like a rolling stone: The changing use of premodifying present participles. English Language and Linguistics 9. 131–156. Fanego, Teresa. 1996. The development of gerunds as objects of subject-control verbs in English (1400–1760). Diachronica 13. 29–62. Fanego, Teresa. 1998. Developments in argument linking in early Modern English gerund phrases. English Language and Linguistics 2. 87–119.

Subordination, or the permanent allure of the “adjacent possible” 

 19

Fanego, Teresa. 2004a. On reanalysis and actualization in syntactic change: the rise and development of English verbal gerunds. Diachronica 21 (1). 5–55. Fanego, Teresa 2004b. Some strategies for coding sentential subjects in English: from exaptation to grammaticalization. Studies in Language 28 (2). 321–361. Fanego, Teresa 2004c. Is Cognitive Grammar a usage-based model? Towards a realistic account of English sentential complements.Miscelánea. A Journal of English and American Studies 29: 23–58. Special Issue on Language and Linguistics containing the plenary lectures from ESSE7 (Seventh International Conference of the European Society for the Study of English), University of Zaragoza, 8–12 September 2004. Fanego, Teresa. 2016. The Great Complement Shift revisited: The constructionalization of ACC-ing gerundives. Functions of Language 23 (1). 84–119 Givón, Talmy. 2009. The genesis of syntactic complexity. Amsterdam: John Benjamins. Goldberg, Adele.E. 1995. Constructions: A Construction Grammar Approach to Argument Structure. Chicago: Chicago University Press. Haiman, John. 1974. Concessives, conditionals, and verbs of volition. Foundations of Language, 11 (3). 341–359. Harris, Alice C. 2003. Cross-linguistic perspectives on syntactic change. In Brian D. Joseph and Richard D. Janda, (eds.), The Handbook of Historical Linguistics, 529–551. Oxford: Oxford Blackwell Publishing. Harris, Alice C. and Lyle Campbell. 1995. Historical syntax in cross-linguistic perspective. Cambridge: Cambridge University Press. Heine, Bernd and Tania Kuteva. 2007. The genesis of grammar. Oxford: Oxford University Press. Hogan, Michael J. and Lisa A. Hogan. 1998. ‘Impertinent’ by sentences in American writing: a cognitive motivation. Word 49. 181–203. Hopper, Paul, and Elizabeth Closs Traugott. 2003. Grammaticalization. Cambridge: Cambridge University Press. Huddleston, Rodney and Geoffrey K. Pullum. 2002. The Cambridge Grammar of the English Language. Cambridge: Cambridge University Press. Johnson, Steven. 2010. The genius of the tinkerer. The Wall Street Journal. (https://www.wsj. com/articles/SB10001424052748703989304575503730101860838, accessed 1 May 2017). Kauffman, Stuart A. 2003. The adjacent possible: A talk with Stuart A. Kauffman (November 9, 2003). Edge Foundation (https://www.edge.org/conversation/the-adjacent-possible), accessed 13 September 2016. Kloos, Heidi & Van Orden, Guy 2009. Soft-Assembled Mechanisms for the Grand Theory.In John P. Spencer, Michael Thomas & James McClelland (eds), Toward a New Grand Theory of Development! Connectionism and Dynamic Systems Theory Reconsidered, 253–267. Oxford: Oxford University Press. König, Ekkehard 1985. On the history of concessive connectives in English. Diachronic and synchronic evidence. Lingua 66. 1–19. König, Ekkehard, & van der Auwera, Johan 1988. Clause integration in German and Dutch conditionals, concessive conditionals, and concessives. In J. Haiman, & S. A. Thompson (eds.), Clause Combining in Grammar and Discourse, 101–134. Amsterdam, Philadelphia: John Benjamins. Langacker, Ronald. 1991. Foundations of Cognitive Grammar, Vol. 2: Descriptive application. Stanford: Stanford University Press. Langacker, Ronald. 1992. Prepositions as grammatical(izing) elements. Leuvense Bijdragen 81. 287–30

20 

 Carlos Acuña-Fariña, Ignacio Palacios-Martínez and Elena Seoane

Lass, Roger. 1990. How to do things with junk: exaptation in language evolution. Journal of Linguistics 26. 79–102. Lass, Roger. 1997. Historical linguistics and language change. Cambridge: Cambridge University Press. Lehmann, Christian. 1988. Towards a typology of clause linkage. In John Haiman and Sandra Sandra Thompson (eds.), Clause Combining in Grammar and Discourse, 181–225. Amsterdam: Benjamins. Los, Bettelou. 2005. The rise of the to-infinitive. Oxford: Oxford University Press. Miller, D. Gary. 2002. Nonfinite Structures in Theory and Change. Oxford: Oxford University Press. Ogura, Mieko & William S.-Y. Wang. 1996. Snowball effects in lexical diffusion: The development of -s in the third person singular present indicative in English. In Derek Britton (ed.), English historical linguistics 1994: Papers from the 8th International Conference on English Historical Linguistics, 119–141. Amsterdam: Benjamins. Petré, Peter. 2012. General productivity: How become waxed and wax became a copula. Cognitive Linguistics 23 (1). 27–65. Petré, Peter. 2014. Constructions and environments: Copular, passive, and related constructions in Old and Middle English. Oxford: Oxford University Press. Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech and Jan Svartvik. 1985. A Comprehensive Grammar of the English Language. London and New York: Longman. Rączaszek-Leonardi, J. 2012. Language as a System of Replicable Constraints. In RączaszekLeonardi, J & Pattee, H. H. (eds.), Laws, Languages and Life, 295–333. Dordrecht: Springer. Rączaszek-Leonardi, J. 2014. Multiple systems and multiple time scales of language dynamics: Coping with complexity. Cybernetics and Human Knowing, Vol. 21 (1–2). Rohdenburg, Günter. 2006. The role of functional constraints in the evolution of the English complementation system. In Christiane Dalton-Puffer, Nikolaus Ritt, Herbert Schendl & Dieter Kastovsky (eds.), Syntax, style and grammatical norms: English from 1500–2000, 143–166. Frankfurt: Lang. Rudanko, Juhani. 2017. Infinitives and Gerunds in Recent English. London: Palgrave Macmillan/ Springer. Schneider, Edgar W. 2007. Postcolonial English: Varieties around the World. Cambridge: Cambridge University Press. Timberlake, Alan. 1977. Reanalysis and actualization in syntactic change. In: Charles N. Li, (ed.), Mechanisms of syntactic change, 141–177. Austin and London: University of Texas Press. Traugott, Elizabeth Closs. 1989. On the rise of epistemic meanings in English: An example of subjectification in semantic change. Language 65. 31–55. Traugott, Elizabeth Closs. 2007. The concepts of constructional mismatch and type-shifting from the perspective of grammaticalization. Cognitive Linguistics 18 (4). 523–557. Traugott, Elizabeth Closs & Graeme Trousdale. 2013. Constructionalization and Constructional Changes. Oxford: Oxford University Press. Traugott, Elizabeth Closs. 2008. The grammaticalization of NP of NP patterns. In Alexander Bergs & Gabriele Diewald (eds.), Constructions and language change, 23–46. Berlin: Mouton de Gruyter. Van de Velde, Freek, De Smet, Hendrik & Lobke Ghesquière. 2013. On multiple source constructions in language change. Special issue of Studies in language 37 (3). Amsterdam: John Benjamins. Van Orden, Guy, Holden, John G. & Turvey, Michael. T. 2005. Self-organization of cognitive performance. Journal of Experimental Psychology: General 132. 331–350.

Subordination, or the permanent allure of the “adjacent possible” 

 21

Verstraete, Jean-Christophe. 2007. Rethinking the Coordinate-Subordinate Dichotomy. Interpersonal Grammar and the Analysis of Adverbial Clauses in English (Topics in English Linguistics 55). Berlin: Mouton. Vosberg, Uwe. 2003. The Role of Extractions and HorrorAequi in the Evolution of -ing Complements in Modern English. In Günter Rohdenburg and Britta Mondorf (eds.), Determinants of Grammatical Variation in English, 305–327. Berlin: Mouton de Gruyter. Vosberg, Uwe. 2006. Die Große Komplementverschiebung: Außersemantische Ein üsse auf die Entwicklung satzwertiger Ergänzungen im Neuenglischen. Tübingen: Narr. Winter, Eugene. 1982.Towards a Contextual Grammar of English. The Clause and its Place in the Definition of the Sentence. London: Allen and Unwin.

Part I: Diachronic perspectives on subordination

Hendrik De Smet

From flying Sancho to swooning Altisidora: The changing use of premodifying present participles in three English translations of Cervantes’ Don Quijote Abstract: Over the Modern period, English has come to make increasing use of situation-oriented present participles to structure narrative texts. Situation-oriented participles are premodifying participles that function as compressed subordinate clauses, relating one event to another. For example, in He hauled the sobbing boy to his feet (British National Corpus, BNC), the event denoted by the participle (sobbing) is construed as backgrounded and approximately simultaneous to the event of the higher clause (hauled). Because situation-oriented participles are used in a specifically clause-like way, their rise can be thought of as a departure from the adjectival character of premodifying participles, and by that token as an instance of functional verbalization. In this chapter, the rise of situation-oriented participles is addressed through a close analysis of their use in three successive English translations of Cervantes’ Don Quijote, roughly representative of Early Modern English, Late Modern English and Present-day English. As in untranslated fiction, the use of situation-oriented participles increases over the three translations. Closer quantitative and qualitative analysis reveals that situation-oriented participles are employed with growing freedom in the successive translations. Situation-oriented participles are increasingly used in contexts where the Spanish original offers a different construal of event relations. Moreover, there are indications that the simultaneity requirement, to which their use is subject, is loosened in Present-day English. This confirms that over time situation-oriented participles became more prominent as a grammatical resource available to English authors and translators.

1 Introduction In terms of form, grammars offer different ways of linking clauses. Functionally, these code different construals of the relations between States of Affairs (SoAs). Subordination is, typically, the grammatical strategy of choice to construe the Hendrik De Smet, KU Leuven https://doi.org/10.1515/9783110583571-002

26 

 Hendrik De Smet

relation between SoAs as asymmetric. This asymmetry may mean that the profile of one SoA overrides that of the other, that one SoA is foregrounded while the other is backgrounded, or that one is asserted while the other is non-asserted (Langacker 1991: 436; Cristofaro 2003: 30–33; see also introduction, this volume). While grammaticalization theory holds that subordination typically develops from paratactic source structures (Heine & Kuteva 2007; Givón 2009), Anglicists know of an alternative route towards subordination: the syntactic reorganization of phrases into clauses. This process is familiar to Anglicists because English boasts a prominent example in the verbalization of the gerund in this language. The English gerund developed from a fully nominal structure, as in (1a) (where scotungum takes dative plural case and is premodified by an adjective), into a non-finite clause, as in (1b) (where feling takes a direct object as well as the adverbial negator not). (1) a. þa wunda þe þa wælhreowan hæþenan mid gelomum the wounds that the barbarous heathens  with  frequent scotungum on his lice macodon (ÆLS(Edmund)181) shootings on his body made ‘the wounds that the barbarous heathens had made on his body with numerous shots’ b. Al our pes … is raþer to be sette in meke suffryng þan in not feling contrarieties. (c1430, OED) ‘All our peace of heart is rather in being put in meek suffering than in not feeling any opposition.’ In fact, changes of this kind can be reconstructed for many languages (Disterheft 1981). What makes the verbalization of the English gerund particularly interesting is that the change happened recently enough that it can be fully traced through the written historical record, starting in Middle English and continuing into the Modern period. As a result, it is possible to examine what triggered the change (Jack 1988; Fanego 2004), how the change unfolded over time (Tajima 1985; Fanego 1998), what functional pressures were in operation (Fanego 2004; De Smet 2008; Fonteyn 2016), and how it affected other clause types (Fanego 1996; Vosberg 2006). One insight that can be derived from the study of the English gerund is that verbalization processes do not only affect the formal side of a construction. The formal changes that characterize verbalization go hand in hand with functional changes. Complex interactions between form and function were first hinted at by Fanego (2004: 17), who suggests that in Middle English formally nominal gerunds could be used with subordinate clause-like functionality. Nominal

From flying Sancho to swooning Altisidora 

 27

gerunds, it seems, anticipated the formal changes that were to follow. Taking this cue, Fonteyn (2016) re-examines the history of the English gerund, focusing on such parameters that might be taken to distinguish clause-like from phrase-like behavior on a functional level. She not only finds that verbal gerunds came to take up more clause-like functions over time but also confirms that the functional changes sometimes also affected formally nominal gerunds, especially during the earlier stages of change. Thus, the gerund in (2), though nominal in form, behaves rather like a non-finite subordinate clause, for instance in being controlled by the higher clause subject and in being temporally integrated in the higher clause. Later, such clause-like nominal gerunds dropped out of use again. (2) And after that he was gon from hem / they made to take vp the ancres & to hale vp their saylles, wher-in þe wynde entred that had them soone ferre from the lande out of syght / & toke the hyghe see assone as they myght, sayllyng by the costes of many a strange regyon wythout fyndyng of eny aduenture that is to be recounted (1489, IMEPCS) ‘And after he had left them, they began to lift the anchors and haul up their sails, which were soon caught by the wind, which made them travel away from the land and out of sight, and they took the high sea as soon as they could, sailing along the coasts of many strange lands without encountering any adventure that is worth telling.’ Verbalization, then, may involve an intricate give-and-take between both formal and functional changes, whereby the functional changes are not necessarily restricted to the most clause-like forms. In other words, there is a functional side to verbalization processes that is to some extent independent of any formal changes. While the development of the gerund is the most spectacular change of its kind in the recent history of English, it may not be the only one. De Smet & Vancayzeele (2014) describe changing usage preferences for English premodifying present participles that suggest another less conspicuous verbalization process. Generally, premodifying present participles can take up three major functions. First, identifying present participles, such as remaining in (3a), aid identification of the nominal referent, typically by signaling how the referent relates to the speech situation or to other identifiable referents in the discourse. In this case, remaining sets up a phoric relation between two straps of a watch, one of which is the focus of attention now, while the other has been the focus of attention at some point in the preceding discourse. Second, type-oriented present participles, such as floating in (3b), refine the type-description provided by the head noun by specifying an intrinsic feature of its referent. Thus, floating is used to specify a type

28 

 Hendrik De Smet

of restaurant. Finally, situation-oriented present participles, such as exploding in (3c), describe how the referent of the head noun engages in the SoA expressed by the higher clause. Exploding here describes an SoA that takes place at some point during what is described by the higher clause. Situation-oriented participles do not describe an intrinsic property of the head noun referent, but one that holds only for the approximate duration of a contextually given situation. (3) a. Philip fiddled with the remaining strap of his watch. (BNC) b. The floating restaurant had the same equipment that every other restaurant had. (BNC) c. She had run from the exploding craft and collapsed. (BNC) Situation-oriented participles as in (3c) effectively function as maximally compressed subordinate clauses, since they describe an event that happens approximately within the time-frame evoked by the higher clause. In doing so, they construe the SoA coded by the present participle as backgrounded and non-asserted. As De Smet & Vancayzeele (2014) show, situation-oriented participles became increasingly common during the Late Modern and Present-day English periods, at least in narrative fiction. This development can be interpreted as another example of functional verbalization – in this case one that shows little or no evidence of any accompanying formal change. Such a development makes sense within the bigger picture of the history of English. It illustrates the increasing reliance on the prenominal slots of the noun phrase to accommodate all sorts of information that used to be coded outside the noun phrase (Biber & Gray 2011) and also testifies to the growing fondness of English for backgrounding constructions (Los 2009; Petré 2014).

2 Goal and methodology In this contribution, I will revisit the rise of situation-oriented present participles in English, drawing on new data that complement the dataset used by De Smet & Vancayzeele (2014). Where De Smet & Vancayzeele relied on traditional corpus data, using the Corpus of Late Modern English Texts (CLMET) and the Freiburg-LOB Corpus (F-LOB) (see below), this follow-up study is based on a small translation corpus, consisting of three English translations of Cervantes’ Don Quijote as well as the Spanish original. Cervantes’ Don Quijote (DQ-S) was published in two parts in 1605 and 1615. The first English translation is that of Thomas Shelton (DQ-E-16), also published in two parts in 1612 and 1620, closely following

From flying Sancho to swooning Altisidora 

 29

the original. The second translation included in the corpus is by John Ormsby and was published in 1885 (DQ-E-18). The third translation is by Edith Grossman and was published in 2003 (DQ-E-20).1 In this way, the English translations in the corpus correspond roughly to Early Modern English, Late Modern English and Present-day English, respectively. The English component of the corpus will be referred to here as DQ-E. The full corpus contains just over 1.6 million words, with each version of Don Quijote contributing slightly over 400,000 words. I have made no systematic attempt to establish how the three translations relate to the source text or to one another. Having worked with them, it appears that they are all very faithful to the Spanish original. They also appear to have been made independently of one another, with none of the passages examined for this study showing signs of having been copied from an earlier translation. While at present such observations are merely impressionistic, the reader can judge for him or herself from the parallel examples quoted in Section 5 below. Though small, DQ-E has two distinct advantages. First, the corpus keeps genre constant over time. This is often difficult in traditional diachronic corpus research because even a genre-balanced or genre-specific corpus consisting of untranslated texts cannot keep genre completely stable, as genres themselves are moving targets – in fact, this is one of the problems found by De Smet & Vancayzeele (2014). DQ-E avoids this problem to a great extent because all texts in the corpus translate exactly the same source text. Second, DQ-E can provide insights into how translators at different stages of the language exploit their linguistic repertoires to convey roughly identical messages. Because communicative intent is kept constant, whatever is found to change in the corpus is likely to reflect changing preferences in how experience is verbalized. Known diachronic changes in the function of specific constructions can be expected to be seen as changing translation practices. Also, by keeping communicative intent constant, the corpus reveals functional equivalences between expressions across texts, providing valuable insight into the semasiological range of an expression at any given time. This, too, is far harder to achieve using traditional corpus methods. Drawing from DQ-E, premodifying present participles have been collected using the regular expression “\b(an?|the) \w+ing\b” as a search string, retrieving from the three English translations in the corpus any forms ending in -ing and preceded by a definite or indefinite article. This is the strategy used by De Smet &

1  Three of the four texts in the corpus are in the public domain and have, for the purposes of this paper, been downloaded from online text archives. DQ-S and DQ-E was downloaded from Project Gutenberg. DQ-E-16 was downloaded from Early English Books Online. DQ-E-20 is not in the public domain, it has been scanned from the printed book, which is currently still in print.

30 

 Hendrik De Smet

Vancayzeele (2014). The query obviously has imperfect recall, missing a substantial subset of premodifying participles, but achieves precision at a workable level. As a next step, situation-oriented participles were identified and for each attested example the corresponding passages were collected in the other translations as well as in the Spanish original. The result is a small set of situationoriented participles in each of the English translations, along with the alternative renditions of the same passages in each of the two other translations. What these data reveal is discussed in detail in Sections 4 and 5 below, focusing first on general quantitative trends, and then looking at the use of situation-oriented present participles with a more fine-grained qualitative analysis. The qualitative analysis in particular confirms the expectation that situation-oriented present participles gained in prominence as a linguistic resource available to speakers of English. Before turning to the results, however, some preliminary observations are in order on the source language, Cervantes’ Spanish.

3 The source language Cervantes’ Spanish does not have a structure that would very systematically correspond to an English premodifying present participle. The reflex of the Latin present participle is no longer productive and survives only in more or less isolated lexical items. Some typical examples from Cervantes’ Don Quijote include ardiente ‘burning’, bastante ‘enough’, flamante ‘brilliant’, semejante ‘similar’, and andante in the phrase caballero andante ‘wandering knight’. On the whole, recourse to this pattern is not frequent, and because some or even most types have developed specialized meanings, translation by an adjective rather than a present participle is common and indeed is sometimes the only possibility. The Spanish structure that comes closest to functioning like an English present participle is the gerund, illustrated in (4). As the examples show, the Spanish gerund is typically used adverbially, as in (4a), as complement to the preposition en, as in (4b), or in a progressive construction with estar, as in (4c). (4) a. se le pasaban las noches leyendo de claro en claro (DQ-S) ‘his nights were spent reading from sunset to sunrise’ b. Llega Cipión a África, tropieza en saltando en tierra (DQ-S) ‘when Scipio arrived in Africa, he stumbled as he leaped ashore’ c. al cual estaban desarmando las doncellas (DQ-S) ‘who the maidens were relieving of his armour’

From flying Sancho to swooning Altisidora 

 31

While these gerunds often allow translation by an English present participle, they do not occur as noun modifiers. Therefore, they too fail to provide a likely source for premodifying present participles in the English translations. In sum, there is relatively little in the Spanish source text that would have given its English translators occasion to use premodifying present participles. Not surprisingly, then, premodifying present participles are underrepresented in the English translations, compared to their use in untranslated English narrative fiction (see Section 4). This means that the participles that do appear correspond to a heterogeneous mix of Spanish source patterns. It also means that where premodifying participles occur in the English data, it is likely that they have been chosen despite the original, as a more idiomatic alternative to a more literal translation from Spanish.

4 Changing frequencies Premodifying present participles gain in frequency in the successive English translations of Cervantes’ Don Quijote. This is shown in Figure 1, which gives normalized frequencies (per 100,000 words) for premodifying present participles in DQ-E (all types included). A steady increase is also seen in the subset of situation-oriented participles. Premodifying participles

Situation‐oriented participles

35 30 25 20 15 10 5 0 1650

1700

1750

1800

1850

1900

1950

2000

Figure 1: Premodifying present participles in DQ-E (frequencies per 100,000 words).

While this overall trend is more or less as expected, the figures should not be given too much weight. They differ in various respects from those reported in De

32 

 Hendrik De Smet

Smet & Vancayzeele (2014), who collected their data from the narrative fiction components of CLMET (covering the period 1710–1920) and F-LOB (sampling texts published in 1991). Most importantly, present participles are about two to three times less frequent in DQ-E than in the CLMET and F-LOB data, and situation-oriented participles in particular take up a much smaller share in DQ-E than they do in CLMET and F-LOB.2 Given the grammar of the source language from which the DQ-E texts were translated, this is unsurprising, as premodifying present participles are comparatively uncommon in DQ-S (see Section 3 above). Generally, features that are unique to the target language will tend to be underrepresented in translation (Tirkkonen-Condit 2004), and this is reflected in the use of premodifying participles in DQ-E. Because it is influenced by its Spanish source text, the DQ-E data is not the best representative of written English fiction when it comes to establishing quantitative trends. Nevertheless, a closer look is instructive. Focusing on situation-oriented participles, Figure 2 shows the distribution of the pattern over the three translations in DQ-E. For each text, the diagram visualizes how many situation-oriented premodifying participles occur and how many of them correspond to the same structure in either of the other two texts. This reveals an interesting difference between the earliest translation, DQ-E-16, and the two later ones, DQ-E-18 and DQ-E-20. Five out of six situation-oriented participles (83%) in DQ-E-16 are also translated by situation-oriented participles in one or both of the other two translations. By contrast, for DQ-E-18 and especially DQ-E-20 the bulk of attested situation-oriented participles translate passages that are rendered by some other means in the two other translations. Accordingly, only six out of 14 situation-oriented participles in DQ-E-18 (43%) correspond to a situation-oriented participle in one of the other translations, and for DQ-E-20 the proportion is even lower, at four out of 18 situation-oriented participles (22%). Because the corpus is small, these numbers are modest, but they are suggestive. Specifically, the distributional pattern in Figure 2 points to a functional expansion of sorts. There appear to be passages in the Spanish original that are quite consistently translated by a situation-oriented participle. These passages can be assumed to more or less require a situation-oriented participle in English, and it is therefore striking that the situation-oriented participles in DQ-E-16 nearly

2 It should also be observed that the steady increase seen in the translations is not completely in line with the developments as seen in CLMET and F-LOB. In the latter data set, premodifying participles in general are found to have been on the decline since the early nineteenth century, while the subset of situation-oriented participles increased in frequency throughout the Late Modern period but then remained more or less stable between Late Modern and Present-day English.

From flying Sancho to swooning Altisidora 

 33

DQ–E–16 1 2 3 1 8 DQ–E–18

14

DQ–E–20

Figure 2: Situation-oriented participles in DQ-E-16, DQ-E-18 and DQ-E-20.

all belong to this group. By contrast, situation-oriented participles appear to be used more freely in the later translations, DQ-E-18 and DQ-E-20, where they also occur in contexts that call for them less strongly. This is what one would find if situation-oriented participles became a more dominant grammatical pattern, to which translators increasingly took recourse, even if the passage they were translating allowed other renditions as well. Thus, the distribution in Figure 2 suggests that the translator of DQ-E-16 resorted to situation-oriented participles only when he had to, while the other two translators used the pattern not only where they had to, but also where they could. As argued in the following section, qualitative analysis lends support to this interpretation.

5 Changing functions Situation-oriented participles construe a process as backgrounded but roughly simultaneous to the more salient event described by their superordinate clause. In what follows, I will argue that through the successive translations in DQ-E this construal of two events is selected with increasing freedom. This is reflected in the fact that the successive translations employ situation-oriented participles in contexts where they are increasingly unfaithful to the Spanish original. To see this, let us start by considering the situation-oriented participles in DQ-E-16. A characteristic example is (4). The Spanish text has a structure that is hard to render literally in English. An agentive nominalization volador ‘flyer’ is used as a premodifier to Sancho. However, the structure here evokes a backgrounded process that unfolds simultaneously with the situation in the main

34 

 Hendrik De Smet

clause, which is essentially what an English situation-oriented participle does. Accordingly, DQ-E-16 selects a situation-oriented participle, flying, and DQ-E-18 and DQ-E-20 opt for the same solution. (4) a. mas no por esto cesaban ellos de su risa y de su obra, ni el volador Sancho dejaba sus quejas (DQ-S) b. yet did not they cease from their laughter and labour; nor the flying Sancho from his complaints (DQ-E-16) c. they, however, did not stay their laughter or their work for this, nor did the flying Sancho cease his lamentations (DQ-E-18) d. but this did not stop them from laughing and doing their work, nor did the flying Sancho leave off his complaints (DQ-E-20) Similarly, in (5) DQ-S itself has a situation-oriented present participle, which is rare in the Spanish source text. DQ-E-16 follows suit, as do DQ-E-18 and DQ-E-20. (5) a. tendido sobre la ardiente arena, envía sus quejas al piadoso cielo (DQ-S) b. stretcht upon the burning sand, [he doth] breathe his pittifull complaints to Heaven (DQ-E-16) c. stretched on the burning sand in the full heat of the sultry summer noontide, [he] makes his appeal to the compassionate heavens (DQ-E-18) d. lying on the burning sand, he sends his complaints up to merciful heaven (DQ-E-20) Example (6) has yet another Spanish source construction, that is, a relative clause que caminaban ‘that walked’. Because English tends to avoid finite intransitive verbs without any adjunct-like modifiers (Quirk et al. 1985: 506), rendering the original literally, by using a relative clause, would be somewhat unidiomatic in English. As it is expressed by a subordinate clause, however, the event in the relative clause is backgrounded, and it takes place simultaneously to the action in the higher clause. A situation-oriented participle is therefore an obvious solution, and is used in DQ-E-16, as well as in DQ-E-18 and DQ-E-20. (6) a. apartándose los dos a un lado del camino, tornaron a mirar atentamente lo que aquello de aquellas lumbres que caminaban podía ser (DQ-S) b. so departing somewhat out of the way, they began again to view earnestly what that of the travailling lights might be (DQ-E-16) c. the two retiring to one side of the road set themselves to observe closely what all these moving lights might be (DQ-E-18) d. the two of them moved to the side of the road and began again to look closely to see what those traveling lights might be (DQ-E-20)

From flying Sancho to swooning Altisidora 

 35

Examples (4), (5) and (6) all represent cases where a situation-oriented participle was a natural choice for the English translators. The other situation-oriented participles in DQ-E-16 are of a similar kind, confirming the idea that situationoriented participles are only resorted to in this earliest translation when their choice is strongly motivated. The only surprise is example (7). DQ-E-16 here translates Spanish el orgullo as the swelling Stomack, where the other two translations simply opt for the noun pride. (7) a. en aquel día, […] donde quedó el orgullo y soberbia otomana quebrantada (DQ-S) b. in that very day […], wherein the swelling Stomack, and Ottomanicall pride was broken (DQ-E-16) c. on that day, […] on which the Ottoman pride and arrogance were broken (DQ-E-18) d. on that day, […] when Ottoman pride and arrogance were shattered (DQ-E-20) Swelling stomach was an idiomatic expression in Early Modern English, associated with the meaning “pride”, as the examples in (8) show. (8) a. he could not moderate the aspiring pride of his swelling stomacke (1577, Early English Books Online, EEBO) b. these [stories] were sufficient to feare any christiãs hert, willing to subdue his stubberne & to muche swellynge stomacke againste the wyll of the Lorde (1548, EEBO) In principle, swelling can be analyzed as a situation-oriented participle; on a situation-oriented reading, the process of ‘swelling’ is interrupted by the process denoted by the higher clause verb was broken, so the two SoAs unfold in a single time-frame (if not in a single mental space) and are approximately simultaneous. While it is unexpected to find a situation-oriented participle in DQ-E-16 to translate a passage without at least the implication of two simultaneous events, the choice for the participle has its motivation, being part and parcel of the lexical choice for the idiom swelling stomach. Presumably, the conjunction of near-synonyms, orgullo and soberbia, in the Spanish original, simply led the translator to look for a similar pair of synonyms in English. Given the idiomatic character of the phrase, the processual construal of swelling in (7) may not have been very prominent, as is also suggested by the somewhat clumsy mixing of metaphors. As a counterexample to the general trend, then, (7) is not particularly

36 

 Hendrik De Smet

strong. Overall, situation-oriented participles are used sparingly in DQ-E-16 and only when there is good reason to do so. In DQ-E-18 and DQ-E-20, the use of situation-oriented participles is less constrained. The pattern regularly occurs in contexts where the Spanish original does not have a corresponding backgrounded event. Example (9) is typical. DQ-E-18 translates el medroso villano as the trembling clown (clown being used in its now obsolete sense ‘peasant, rustic’, cf. Oxford English Dictionary, OED, s.v. clown n. 1). Trembling metonymically evokes the fear denoted by Spanish medroso ‘fearful’, but in doing so it also adds a backgrounded SoA that takes place simultaneously to the higher clause SoA. (9) a. b. c. d.

Respondió el medroso villano que […] (DQ-S) The fearfull Countryman answered, That […] (DQ-E-16) The trembling clown replied that […] (DQ-E-18) The terrified farmer replied that […] (DQ-E-20)

A similar example, also from DQ-E-18, is given in (10). By translating la sangre y pecho as the bleeding breast, DQ-E-18 essentially renders the meaning of the original but in doing so adds to the single SoA of Claudia’s lying unconscious on Don Vicente’s bloody breast the additional backgrounded one of Don Vicente’s blood rushing forth from his wounds. (10)

a. Apretóle la mano Claudia, y apretósele a ella el corazón, de manera que sobre la sangrey pecho de don Vicente se quedó desmayada, y a él le tomó un mortal parasismo. (DQ-S) b. Claudia wrung his hand, and herself was wrung to the very heart; so that upon Don Vincent’s blood and brest, shee fell into a swound, and hee into a mortall Paroxisme. (DQ-E-16) c. Claudia wrung his hands, and her own heart was so wrung that she lay fainting on the bleeding breast of Don Vicente, whom a death spasm seized the same instant. (DQ-E-18) d. Claudia pressed his hand, and her own heart felt pressed, causing her to faint onto the bloody bosom of Don Vicente, who was shaken by a mortal paroxysm. (DQ-E-20)

Examples of the same type can be found in DQ-E-20. For instance, in (11) DQ-S has la cuitada, which is a nominalized adjective, itself morphologically derived from a noun, cuita ‘sorrow’. The most literal translation would be ‘the sorrowful one’. Because English tends to avoid nominalized adjectives, all three translators seek to supply a proper nominal head. DQ-E-18 opts for the abstract noun anxiety,

From flying Sancho to swooning Altisidora 

 37

while DQ-E-16 and DQ-E-20 stay closer to the original by using the animate noun princess as head, rendering the sense of anxiety by a premodifier. In DQ-E-16 the premodifier is an adjective, carefull, whereas in DQ-E-20 it is a situation-oriented participle. This again imposes a construal with two SoAs on the scene where the original has only one. (11)

a. b. c. d.

consuélase con esto la cuitada (DQ-S) The carefull Princesse will comfort her selfe with this hope (DQ-E-16) her anxiety is thus relieved (DQ-E-18) the suffering princess consoles herself with this (DQ-E-20)

By using situation-oriented participles, the translations in (9c), (10c) and (11d) add ongoing backgrounded SoAs that are not present in the original. These translation choices are not particularly compelling, confirmed by the fact that they are not followed in the other translations. At the same time, they remain relatively faithful translations in that they do not add anything that conflicts with the original. Going a step further, sometimes both translations and original establish a link between two SoAs in the narrative, but the English translators of DQ-E-18 and DQ-E-20 choose a construal that is in conflict with the one chosen in the Spanish text. The most obvious cases are when an English present participle translates a Spanish past participle. For example, in (12) DQ-E-20 uses burning to translate Spanish encendida ‘lighted’. (12)

a. b. c. d.

y, asiendo de una hacha encendida que junto a él estaba (DQ-S) and laying hold of a lighted Torch that was neere him (DQ-E-16) and seizing a lighted torch that stood near him (DQ-E-18) and grasping one of the burning torches that was near him (DQ-E-20)

In (12), the choice for a present participle may actually be made to satisfy a lexical preference for burn instead of light. In (13), however, lexical choice is constant, yet DQ-E-20 again opts for a present participle where Spanish has a past participle, subtly changing the meaning of the passage. (13)

a. Reincorporóse y redújose el renegado con la Iglesia, y, de miembro podrido, volvió limpio y sano con la penitencia y el arrepentimiento. (DQ-S) b. The Runagate was reduc’t and re-encorporated with the Church, and of a rotten member became clean and sound by penance and repentance. (DQ-E-16)

38 

 Hendrik De Smet

c. and the renegade effected his readmission into the body of the Church and was reconciled with it, and from a rotten limb became by penance and repentance a clean and sound one. (DQ-E-18) d. The renegade was reconciled with and reintegrated into the Church, a rotting limb who became cleansed and healthy again through penance and repentance. (DQ-E-20) In these examples the choice for a present participle is certainly not necessitated by the original, which offers an alternative construal of event relations that could have been rendered in English just as easily. Situation-oriented present participles are used more or less gratuitously here, suggesting that they have become more readily available to translators as a linguistic resource. Another indication of this is that DQ-E-20 in particular appears to loosen the semantic constraints on the use of present participles. The most striking example is (14). (14)

a. No se hubo bien apartado, cuando, volviendo en sí la desmayada Altisidora, dijo a su compañera […] (DQ-S) b. hee was no sooner gone, when the dismayed Altisidora comming to her self, said to her companion […] (DQ-E-16) c. He had scarcely withdrawn when Altisidora, recovering from her swoon, said to her companion […] (DQ-E-18) d. No sooner had he gone away than the swooning Altisidora came to her senses and said to her companion […] (DQ-E-20)

The passage in (14) is part of the following narrative sequence: intending to play a trick on Don Quixote, Altisidora pretends to faint as Don Quixote enters the room, a brief exchange follows between Don Quixote and the bystanders, Don Quixote leaves, after which Altisidora, abandoning the charade, resumes normal interaction with her companions. Obviously, Altisidora’s swoon and recovery are not strictly simultaneous. Accordingly, in (13), DQ-S and DQ-E-16 use a past participle, desmayada and dismayed (meaning ‘overwhelmed’ OED s.v. dismay v.1 3) respectively, to construe the swoon as prior to the recovery. DQ-E-18 achieves approximately the same effect by nominalizing the SoA of swooning and marking it as discourse-given, which is consistent with its being earlier in time and therefore prior in the narrative. DQ-E-20, however, renders the same SoA by a present participle. This seems illogical, but it is reminiscent of other uses of present participles in Present-day English where a simultaneity construal is chosen despite the absence of any actual simultaneity. One such example, discussed by De Smet & Heyvaert (2011: 484), is (15).

From flying Sancho to swooning Altisidora 

(15)

 39

The disappearing mayor of the French Riviera resort of Nice, Honore Bailet, 73, made his first public appearance in five weeks yesterday (Google)

In both (14d) and (15), what is simultaneous to the higher clause SoA is not the actual SoA denoted by the present participle (“the swoon” or “the disappearance”, respectively), but its ripple effects and continued relevance within the communicative setting. Simultaneity, here, has become “sloppy” (Declerck 1991: 132–4). A possibly similar example is (16). Again, the context has a straightforward chronological order of events. Heading into the mountains, Don Quixote and Sancho Panza find an abandoned trunk and saddle-cushion. A little later they encounter a man in rags who immediately runs away from them. Don Quixote ponders the events and concludes that the man must be the owner of the abandoned goods, which he believes is confirmed when later they also find a dead mule. In real time, there is a considerable time lapse between the various events, so it is striking that where DQ-S, DQ-E-16 and DQ-E-18 simply mark the man’s fleeing as an SoA belonging to the prior time sphere, DQ-E-20 can still refer to the man in rags as the fleeing man. (16)

a. todo lo cual confirmó en ellos más la sospecha de que aquel que huía era el dueño de la mula y del cojín. (DQ-S) b. all which confirmed more in them the suspition, that hee which fled away was owner of the Mule and Cushion. (DQ-E-16) c. all which still further strengthened their suspicion that he who had fled was the owner of the mule and the saddle-pad. (DQ-E-18) d. which was further confirmation of their suspicion that the fleeing man was the owner of both the mule and the saddle cushion. (DQ-E-20)

This can be interpreted as another example of sloppy simultaneity, with an earlier event being marked as ongoing because it continues to reverberate into the current time of orientation of the narrative.

6 Conclusions Situation-oriented present participles may not be a Modern English innovation, but their use increased over the Modern period, at least in narrative fiction. As such, the change presents a recent case of functional verbalization.

40 

 Hendrik De Smet

In this study, the rise of situation-oriented participles has been studied through the lens of three English translations of Cervantes’ Don Quijote. This approach obscures some aspects of the development, because the translations are of course influenced by the grammar of the Spanish original and not directly representative of ongoing changes in the target language, English. But it also reveals something that is otherwise difficult to see. Over time, situation-oriented participles take an increasingly prominent position in the translators’ repertoire of grammatical structures. Not only do they come to be used more frequently, analysis of the data reveals that they come to be employed more freely. Several pieces of evidence point in this direction. First, some passages in the Spanish Don Quijote are particularly likely to be translated with an English situation-oriented present participle. Passages can be identified as such when they are translated by a situation-oriented participle more than once. The quantitative analysis suggests that it is only for these specific passages that the earliest English translation resorts to situationoriented participles. In contrast, the two later translations also use the pattern elsewhere. Second, qualitative analysis suggests a distinction between more and less faithful uses of situation-oriented participles. The uses in the earliest translation tend to be highly faithful. Situation-oriented participles are used in English to render passages in Spanish that also have a backgrounded SoA which unfolds simultaneously with some other more salient SoA. By contrast, the more recent translations resort to situation-oriented participles more freely, sometimes using them where the original does not have a two-event construal at all, or where the original has a two-event construal but one that would be more faithfully rendered by other grammatical means in English. Third, the most recent translation shows the occasional use of situationoriented present participles with “sloppy” simultaneity. This suggests a functional expansion of the pattern. Not only did the pattern become more favored as a grammatical resource, its functional range expanded to include relations between SoAs that are not strictly simultaneous. In its “sloppy” use, the pattern is exploited to convey temporal and associative proximity between SoAs. These different findings are of course based on a very modest data set, but they corroborate earlier work and in doing so shed new light on previous findings. Specifically, what the data show is how a grammatical resource that grows in frequency also comes to take a more central place in language users’ linguistic repertoires, and comes to be resorted to more freely. This process need not also be accompanied by (subtle) functional changes, but it can be. In the case of situation-oriented present participles, increasing freedom of use reflects

From flying Sancho to swooning Altisidora 

 41

routinization and appears to be accompanied by a functional change understood here as “sloppification”.

Data sources BNC = British National Corpus CLMET = Corpus of Late Modern English Texts EEBO = Early English Books Online F-LOB = Freiburg-LOB Corpus IMEPCS = Innsbruck Middle English Prose Corpus – Sampler OED = Oxford English Dictionary (4th ed.)

References Biber, D. & B. Gray. 2011. Grammar emerging in the noun phrase: The influence of written language use. English Language and Linguistics 15. 223–250. Cristofaro, S. 2003. Subordination. Oxford: Oxford University Press. Declerck, R. 1991. A comprehensive descriptive grammar of English. Tokyo: Kaitakusha. De Smet H. & L. Heyvaert. 2011. The meaning of the English present participle. English Language and Linguistics 15. 473–498. De Smet H. & E. Vancayzeele. 2014. Like a rolling stone: The changing use of premodifying present participles. English Language and Linguistics 19. 131–156. Disterheft, D. 1981. Remarks on the history of the Indo-European infinitive. Folia Linguistica Historica 2. 3–34. Fanego, T. 1998. Developments in argument linking in early Modern English gerund phrases. English Language and Linguistics 2. 87–119. Fanego, T. 1996. The development of gerunds as objects of subject-control verbs in English (1400–1760). Diachronica 13. 29–62. Fanego, T. 2004. On reanalysis and actualization in syntactic change: The rise and development of English verbal gerunds. Diachronica 21. 5–55. Fonteyn, L. 2016. Categoriality in language change: The case of the English gerund. Leuven: University of Leuven unpublished doctoral dissertation. Givón, T. 2009. The genesis of syntactic complexity. Amsterdam: John Benjamins. Heine, B. & T. Kuteva. 2007. The genesis of grammar. Oxford: Oxford University Press. Jack, G. 1988. The origins of the English gerund. Nowele 12. 15–75. Langacker, R. 1991. Foundations of Cognitive Grammar, Vol. 2: Descriptive application. Stanford: Stanford University Press. Los, B. 2009. The consequences of the loss of verb-second in English: Information structure and syntax in interaction. English Language and Linguistics 13. 97–125.

42 

 Hendrik De Smet

Petré, P. 2014. Constructions and environments: Copular, passive, and related constructions in Old and Middle English. Oxford: Oxford University Press. Quirk, R., S. Greenbaum, G. Leech & J. Svartvik. 1985. A comprehensive grammar of the English language. London: Longman. Tajima, M. 1985.The syntactic development of the gerund in Middle English. Tokyo: Nan’un-do. Tirkkonen-Condit, S. 2004. Unique items: Over- or under-represented in translated language? In A. Mauranen & P. Kujamäki (eds.), Translation universals: Do they exist?, 177–184. Amsterdam: Benjamins. Vosberg, U. 2006. Die Große Komplementverschiebung: Außer-semantische Einflüsse auf die Entwicklung satzwertiger Ergänzungen im Neuenglischen. Tübingen: Narr.

Alexander Bergs

Because science! Notes on a variable conjunction Abstract: The American Dialect Society (ADS) named because Word of the Year 2013 as this  “very old word […] exploded with new grammatical possibilities”. What they referred to was the fact that in Present-day English, because is not necessarily followed by of or a full clause, as in (1) and (2), respectively, but shows a surprisingly large variety of different complements, as in (3)–(5). (1) It’s tough because of your language (Corpus of Contemporary American English, COCA, Journal of Adolescent and Adult Literacy Feb 2014) (2) And so because he’s so troubled she calls in her minister to talk with her brother. (COCA, NPR 2012, Margaret: Inside the Fall Of A Teenager) (3) But motivation alone does not assure success “Because circumstances.” (COCA 1996, International gifted women: Developing a critical human resource) (4) Eat veggie burger. Finish this one, but only because famished. (COCA 2005, Going for the daily double) (5) taeyeon is so lucky because goddamnit (Bohmann 2016) Even though these grammatical possibilities are perhaps not as new as the ADS might think, few studies have thus far looked into this phenomenon (Hirose 1991; Bailey 2012, 2014; Pullum 2014; Kanetani 2015; Bohmann 2016). These have mostly concentrated on matters of syntactic theory or register distribution. It was claimed that the new constructions originate from computer-mediated communication, or at least are very typical of it. The current paper considers the structures themselves and the uses of these “new” constructions in spoken, written, and electronic media. The analysis adopts an approach based largely on Construction Grammar (as does Kanetani 2015). One of the central theoretical aims of this paper is to develop a model of variation and change on the syntactic level from a Note: This paper has been greatly improved following valuable comments and questions by the anonymous referees and Elizabeth Traugott. I am very grateful for their insightful suggestions. All errors are, of course, my own. Alexander Bergs, Osnabrück University https://doi.org/10.1515/9783110583571-003

44 

 Alexander Bergs

constructional point of view, using the development of because as an example. It will be concerned with a micro-perspective on how and why speakers manipulate constructions and the constructicon (see Section 2), and how this may ultimately lead to syntactic change on the macro-level of the speech community. It will add to the current debate concerning because by discussing economy, subjectivity, and viewpoint as important factors in the development and current use of the “new” because.

1 Because when? Quirk et al. (1985: 669, 998) describe because of as either a complex preposition, with the first part because being an adverb and the second part of a preposition, or, alternatively, as a (simple) subordinating conjunction because followed by a finite clause, as in (1) and (2) above. Similarly, Pullum & Huddleston (2002: 624–25) note that because of may be seen as a “composite” preposition, where of + NP may be replaced by a content clause (e.g. because I can’t find it). Neither Quirk et al. nor Pullum & Huddleston mention the possibility that because may be followed by a simple NP or a simple AdjP, as in (3) and (4) above. In 2014, The American Dialect Society (ADS) made the surprising choice of because as their Word of the Year 2013.1 Ben Zimmer, Chair of the New Words committee, explained that [t]his past year, the very old word because exploded with new grammatical possibilities in informal online use, […] No longer does because have to be followed by of or a full clause. Now one often sees tersely worded rationales like “because science” or “because reasons”. You might not go to a party “because tired”. As one supporter put it,  because  should be Word of the Year “because useful!” (American Dialect Society).

This, however, does not seem to be strictly accurate, for two reasons. First, the because + AdjP combination can be traced back at least to the early nineteenth century. It can be found in the work of Charlotte Brontë, for example: (6) be sure that there is a canker somewhere, and a canker not the less deeply corroding because concealed. (Charlotte Brontë, Shirley 1849)

1 http://www.americandialect.org/2014/page/4

Because science! Notes on a variable conjunction 

 45

The Corpus of Historical American English (COHA)2 also lists a few examples from the early nineteenth century.3 (7) If Southern white men think it unsafe to permit the negroes to exercise or possess the right of suffrage, let them disfranchise them, and then surrender so much of their congressional representation as is based on the black population of their States, // This would at least be honest, though I think it would be unwise, because unnecessary. BETTER TO GIVE EVERYBODY A FAIR CHANCE. (COHA, 1820, North American Review) (8) A New-Yorker, to prove the right of his city to the first rank, refers to the next census. And a Bostonian, appeals to history, and shows that Boston is first, because oldest. (COHA, 1823, John Neal, Randolph. A Novel)  While this combination is relatively rare overall (for example, there are some 69 occurrences in COCA between 1810 and 2000), there seems to be a slight peak (with six occurrences) in the early twentieth century. Because + NP combinations seem to be rarer in the history of the language. In COHA only one clear example can be found, from 1940: (9) Because I suppose a lunatic’s a person with a one-track mind. Whereas this? “he made a circular gesture” this is a no-track mind. No-track because infinity-track. It’s the mind of an idiot of genius. (COHA 1940, Harper’s Magazine, Huxley, After many a summer) This means that the claim that because exploded with “new grammatical possibilities” is not entirely correct. Because + AdjP has been around for at least two hundred years, probably more, and because + NP can be found at least from the 1940s onwards. The second inaccuracy in the statement of the ADS has to do with register. Zimmer claims that the “new” because is related to “informal online use”. Neither 2 The Corpus of Historical American English (COHA) contains 400 million words from 1810–2009 (see corpus.bye.edu/corpora.asp). 3 In order to allow for simple, automated searches, the strings (i.e. because followed by an adjective and a period) and (i.e. because followed by a noun and a period) were used. All results were manually checked and corrected. The results were complemented by a second search for and (i.e. because followed by an adjective or a noun, without following period). Unfortunately, these search strings lead to an extremely large number of hits, certainly beyond manual checking. However, it seems that in COHA because + adjective was a lot more common than because + noun.

46 

 Alexander Bergs

the historical data mentioned above nor current examples seem to support this. COCA lists some nine examples of because + NP and twelve of because + AdjP.4 Most of these come from spoken language or (fictional) representations of spoken interaction. There is no reason to believe that these two combinations are restricted to online use only, or that they are triggered by it. However, informal online communication may be a facilitator here and may serve as a breeding ground for the spread of these constructions. We will return to these questions below in discussing questions of style and register. What does seem to be the case, however, is that these constructions appear to become more frequent and more prominent in Present-day English. While COHA (with 400 million words in total over 200 years) only lists a handful of examples in historical English (and almost no combinations with NPs), COCA (currently with 520 million words in total) lists more than twenty occurrences over the last 25 years, nine of which are because + NP constructions. The phrase because science even became an internet meme recently5 and the name of a Youtube channel with Kyle Hill.6 In the 1.9 billion-word corpus of Global Web-based English (GloWbE) because science is also the most common hit (7 occurrences, 3 in the US, 2 in IE, 1 in GB and SG), followed by because duh (3 US, 1 SG), because math (US 3) and because people (US 1, IE 1, SG 1). Similarly, because + AdjP combinations can be found, such as because potential (GB 2, HK 2), because immortal (GB2, IE 1), and because impossible (US1, NZ 1). What we seem to find for both constructions is a significant number of syntactic hapax legomena, that is, new, different tokens, which suggests a certain degree of productivity (cf. Plag 2006). In the next section, these structures and their development will be analyzed from a Construction Grammar perspective. We will first briefly introduce the framework of Construction Grammar (CxG), and then look at the phenomena described above in greater detail.

4 The Corpus of Contemporary American English (COCA) contains 520 million words from 1990– 2015, 20 million words each year from spoken, fiction, popular magazines, newspapers and academic texts (see corpus.bye.edu/corpora.edu). In order to allow for simple, automated searches, the strings (i.e. because followed by an adjective and a period) and (i.e. because followed by a noun and a period) were used. All results were manually checked and corrected. The results were complemented by a second search for and (i.e. because followed by an adjective or a noun, without following period). Unfortunately, these search strings lead to an extremely large number of hits, certainly beyond manual checking. However, when the search was restricted to the spoken section of COCA, the distribution of because + NP and because + AdjP was confirmed. 5 cf. http://www.because-science.org 6 https://www.youtube.com/playlist?list=PLNg1m3Od-GgOshqs9pRZUW7eNUTU-Bsz7

Because science! Notes on a variable conjunction 

 47

2 Because what? CxG begins with the assumption that language does not operate with a “words and rules” system, as mainstream generative approaches do (cf. Pinker 1999). Instead, language consists of constructions only. Constructions are defined as conventionalized form-meaning pairings, and can be found at all levels of linguistic structure, from morphology to syntax, and maybe even beyond (for a comprehensive overview and introduction, see Hoffmann & Trousdale 2016; Hoffmann 2017; for the idea of higher level constructions, see Bergs & Hoffmann 2014, in print). Constructions typically show non-compositionality on their form and/ or meaning side. However, structures which occur with sufficient frequency can also be stored as constructions (cf. Goldberg & Jackendoff 2004: 533). What this means is that the linguistic system is essentially comprised of items based on the key notion of Saussurean signs (words as symbols, i.e. conventional and arbitrary combinations of form and meaning or function) at varying degrees of complexity and abstractness. These include a simple (but abstract) word class, such as Noun, or an abstract (but complex) syntactic pattern such as the passive, or a simple, concrete word (the Saussurean prototype) or complex but concrete idioms.

Abstract Concrete

Simple NOUN green

Complex [Subj BE V-en (PPby)] That’s the way the cookie crumbles.

Constructions have a form side and a meaning/function side. CxG is a comparatively flexible framework in that it allows us to record anything of relevance for the construction on either side. The form side can (but need not) include information on the phonology, morphology, or syntax of the construction in question, while the meaning or function side may include information on its semantics, pragmatics, discourse-functional properties etc. For example, while the going to future construction and the gonna future construction share the same semantic meaning (they both refer to the future) on the function side, they show differences on the form side (different phonology and morphology) and on the function side (they both refer to the future, but differ in their discourse-functional properties, as gonna is marked for informality). These differences can and should be recorded in the constructional properties. Constructions are stored in the mental constructicon, which can be conceptualized in analogy to the mental lexicon, i.e. as structured inventory with links, hierarchies etc. The future construction going to, for example, has formal links to both the full verb go and the preposition to, as well as functional links to go, to, and other future constructions such as will (cf. Bergs 2010).

48 

 Alexander Bergs

Last but not least, while there are a number of different constructional perspectives (see Hoffmann & Trousdale 2016 or Hoffmann 2017 for an overview), the present paper assumes a usage-based approach. This means that the formalism is usage-based and that “motivation” plays a key role in the stipulation of constructions and their relationships. Moreover, the aim is to arrive at a model with psychological plausibility, which is in line with and corresponds to insights from language processing, change, and acquisition, including the role of frequency effects in all these domains. In the following paragraphs, the because + NP/AdjP phenomena discussed in Section 1 above will be discussed from such a usagebased constructional perspective. The linguistic expressions described in Section 1 are comparatively easy to capture from a constructional perspective. At first sight it appears that, on the form side, we find because plus either an NP or an AdjP. From the (limited) data available, and considering data cited in previous studies, it is unclear at this point whether the NP needs to lack a determiner or premodifier, as in (10), where the addition of a modifier such as favorable or of the determiner the leads to questionable results. Note that all examples in COCA or COHA have bare nouns. (10)

An African woman described her people as highly valuing education as the “only passage” to success. But motivation alone does not assure success: “Because [?favorable / ?the] circumstances. I was just lucky, really …” (COCA 1996, Roeper Interview)

Similarly, it is also still unclear whether the AdjP can have premodification, e.g. by very as in (11). Again, no examples with premodifier have been identified in COCA or COHA. (11)

He was not willing to trust the word of the Lord of Shadows. all the earlier attempts might have been preparations for some ingenious masterstroke, deadlier than the rest because [?very] unexpected. (COCA 2004, Sci-Fi)

Bohmann (2016) suggests that one motivation for the use of ‘because X’ may be economy, that is, maximally condensed structures. This would support the idea that NP or AdjP modification is at least rare, as it runs counter to this maximal economy. So, unless such structures can be found, we can assume a bare form for this construction, as in (12). (12)

and

Note that other material can also be found following because.

Because science! Notes on a variable conjunction 

 49

(13)

They turn to see Derek standing behind Maerai, because duh. That is the entire basis of their lives. (GloWbE http://archiveofourown.org/ works/456838)

(14)

I actually enjoy Dilbert, it’s not high art but it hit a chord very precisely and effectively. Not that that speaks in any way to the … behavior evident in this thread. Because wow. (GloWbE http://www.metafilter. com/102472/How-to-Get-a-Real-Education-by-Scott-Adams)

Schnoebelen (2014) presents the following data, based on his analysis of 23,583 tweets. Table 1: Complementation patterns of because in 23,583 tweets (based on Schnoebelen 2014). Part of speech Nouns (people, spoilers) Compressed Clauses (yolo, ilysm)1 Adjectives (ugly, tired) Interjection (sweg, omg)2 Agreement (yeah, no) Pronoun (you, me)

Frequency (word count ≥ 50) 32.02 % 21.78 % 16.04 % 14.71 % 12.97% 2.45 %

 yolo = ‘you only live once’; ilysm = ‘I love you so much’  sweg = swag = ‘stuff we all get’ or ‘sexy, cool, groovy’; omg = ‘oh my god’

1

2

Apparently, the largest group of complementation patterns for because involves nouns (32%), followed by compressed clauses (22%) and then adjectives, interjections, and agreement markers (at about 12–16% each). I am somewhat hesitant to include (reduced, compressed) clauses in this list, as their status is unclear and cannot be discussed at length at this point. As we can tell, the expanded versions are probably not “new” in the same sense as constructions with bare adjectives or nouns: because yolo = because you only live once. This seems to be a question of morphology rather than syntax, and so is not immediately relevant to the present discussion. I would also like to suggest that the categories interjection and agreement can be conflated under the heading of “particle” as they seem to be sufficiently similar in their syntactic behavior. Because plus pronoun occurrences are extremely rare and none could be found in the data for the present study, so they can also be excluded. Taking these findings into consideration, I suggest that the form of the construction can be described in somewhat broader terms, as in (15). (15)

where C2 can be a noun, an adjective, or particle

50 

 Alexander Bergs

This leads us to suspect a micro/meso kind of hierarchy, where the individual because + noun, adjective, particle etc. constructions form micro-constructions, from which the more general meso-construction because X can be abstracted. In the following paragraph, we will take a closer look at the possible functions of this meso-construction. Kanetani (2015) discusses the meaning of the because X construction. On the basis of Sweetser (1990), he investigates the potentially polysemous nature of the construction. Because in its “traditional” context belongs to either content, epistemic, or speech act domains, as in (16a–c). (16)

a. [John come back]1 because [he loves her]2.

b. [John loves her]1, because [he came back]2.

c. [What are you doing tonight]1, because [there is a good movie on]2.

(based on Sweetser 1990: 77)

(16a) relates the two events in a simple, content-oriented way: X1 happened because X2. (16b), on the other hand, not only expresses the causal relationship, but also the speaker’s reasoning behind the conclusion: X2 is the case, therefore X1 is very likely also the case. (16c) does not ultimately establish a strict causal relationship, but rather presents a post-hoc justification and clarification for the speech act: I suggest X1 considering that X2. Interestingly, Kanetani (2015: 66) claims that the because X construction is not triple-polysemous like the regular because construction. He does find a few examples of the epistemic and speech act function in his data, but these are so few, Kanetani claims, that they do not warrant a separate category. In other words, the because X construction mostly deals with the content domain. This is also supported by the data for the present study, where no clear examples for epistemic or speech-act domains can be found. It thus follows that the meaning of the construction can be described as some event or state (expressed in C1), this being the case because of C2. The construction can thus be described as in (17). (17)

BECAUSE X FORM < C1 because C2(N, A, Clause, Particle) >⇔< C1 because C2 >MEANING

As noted above, constructions do not exist in isolation but are stored in a structured mental inventory, the constructicon. Like all other constructions, the because X construction is linked to a number of other constructions. On the one hand, in terms of hierarchy, because X is a meso-construction that interacts with at least three micro-construction, as already pointed out.

Because science! Notes on a variable conjunction 

 51

   because X because N   because A   because particle We can assume that the development and spread of the construction works in a kind of feedback loop as we can also see it in snow cloning.7 In all likelihood, and from what we can tell, the construction started with a few individual tokens of the because + A kind (e.g., because small, because pure, because general). These individual tokens led to the formation of the more general and abstract micro-construction because + A. The availability of this construction in turn fostered and facilitated the creation of new because + A constructs. Because  + N was then formed, probably by analogy, and again started with a few tokens (e.g., because people) before the new and more abstract because + N micro-construction was formed, which again facilitated the formation of more individual constructs of this kind. The same happened for because + particle, although here, due to the lack of data, it is impossible to say whether this came during or after the formation of the even more abstract meso-construction. In any case, as these two or three micro-constructions were established we also see the rise of the more general abstract pattern, the meso-construction because + X. And this, again, leads to more occurrences of the micro-constructions and individual tokens, and probably also the development of new micro-constructions (such as, perhaps, because + particle). Also note that during the development, and especially in Present-day English, we see the formation of some very strong exemplars for the micro-constructions: because science has been discussed above, and other examples include because life or because reasons. UrbanDictionary.com, a popular and influential website for youth language, neologisms, and new linguistic memes, lists about a dozen because + N constructs, including because nope, because shut up, because street, and because Tamara’s boobs.8 In a kind of snow cloning effect these witty and strong exemplars can again lead to the formation of new constructs. On the other hand, the because X construction is also related to a number of other constructions in the constructional network, the constructicon. These include the “regular” causal because construction (John could not come because he was sick; I could not come because of the weather) and the subject because 7 http://languagelog.ldc.upenn.edu/nll/?cat=37 8 Because Tamara’s boobs “refers to the time when Maggie and Kellen wrote ‘YOLO’ on Tamara’s boobs and ‘because YOLO’ became ‘because Tamara’s boobs’. True Story. I killed the neighbors because Tamara’s boobs… Let’s do heroin because Tamara’s boobs…” (urbandictionary.com, s.v. because July 25, 2012).

52 

 Alexander Bergs

construction ([Just] because John is sick doesn’t mean he can’t come). Note that the latter also has a “reduced” alternative, like the because X construction, as in (18). (18)

just because cheap doesnt mean it cant fulfill the purpose that it was designed (https://www.rcgroups.com/forums/showthread.php?316614New-Dragonfly-22a-cp/page4, 23 January 2005)

Constructions like that in (18), however, seem to be particularly rare, at least in this dataset, and cannot be discussed in detail at this point. However, more detailed studies might usefully look into this. The question of how exactly these constructions are related is a difficult one. Kanetani (2015) suggests that “regular” causal because is an instance of because X, since full causal because somehow elaborates on because X, so that John couldn’t come because sick has an elaborate counterpart John couldn’t come because he was sick. In other words, for Kanetani because X is more schematic and instantiates the more concrete full causal because. At the same time, because X is also a proper subpart of causal because, since most because X constructions, Kanetani maintains, are functional and formal subparts of the causal because constructions, as in (19b) for example, which is metonymically related to (19a). Homework as object in this case metonymically captures the action of doing your homework. (19)

a. I can’t go out with you today because I have a lot of homework. b. I can’t go out with you today because homework. (Kanetani 2015: 71)

Formally (syntactically and semantically), this is certainly a very enlightening analysis which highlights the most important relationships between the different constructions. Pragmatically, however, there seem to be differences, which need to be captured in a constructional analysis. This will be the topic of the following section.

3 Because why? The previous analysis has focused on structure and semantic meaning. It has been shown that the because X construction has its own linguistic form, but shares its semantic meaning with the most prominent domain of the “regular” causal because construction. This naturally leads to the question as to whether

Because science! Notes on a variable conjunction 

 53

the two constructions are synonymous or not. The principle of no synonymy, first formulated by Bolinger (1968: 127) and later applied to construction grammar by Goldberg (1995: 67) claims that “[i]f two constructions are syntactically distinct, they must be semantically or pragmatically distinct”. As we have just seen, the because X and the causal because construction are syntactically distinct but semantically similar, in that they both express causal relationships between the two parts of the utterance. So, coming back to the principle of no synonymy, we should be able to find a difference in pragmatics. In this paper, I argue that there are at least two kinds of differences in the pragmatics of the two constructions, and that these can be used to distinguish them. The first difference has to do with style and register. It has been suggested in the literature that because X perhaps originated in computer-mediated communication, or that at least it is a predominantly internet-based form. However, as I have argued above, this is not necessarily the case. The construction itself is a lot older than the internet, and in present-day corpora we find a number of occurrences which have nothing to do with computer-mediated communication. However, most if not all examples in COCA have to do with spoken, informal language. They either come directly from spoken interaction, as in (20), or they occur in representations of spoken interaction, as in (21). Only a handful of examples are found which belong to the written register, as in (22). (20)

But they’re two Northeastern teams, so I’m mad and I’m bitter about it. If you wanted to make people mad, you could shamelessly say, I’m supporting the Patriots because Patriots. So Giants I think will win. GEORGE-STEPHANOPOU# (Off-camera) Okay Giants again. (COCA 2012, ABC this week)

(21)

“We should be very, very concerned. I am terribly concerned,” Cheatham says “Diabetes is a major killer. It doesn’t only cause people to lose their vision, lose their kidneys, lose their limbs, diabetes reduces life expectancy significantly. People die of heart attacks and strokes because diabetes. It is one of the more underlisted causes of death of all causes … With the trend line that we’re on, it’s a terrible epidemic.” (COCA, Magazine 2007)

(22)

Not showy and glittery like the peeling palaces along the Grand Canal, once glamorous beauties now slipping into their dotage, this campo was authentic Venice, surviving because unvisited. It was a blessed remnant of what Venice must have been long ago when Antonio visited his Venetian family, and I have a photo of him as a child in his sailor suit standing at the well in his grandmother’s courtyard. (COCA 2011, Academic Writing)

54 

 Alexander Bergs

There is not enough data in the major corpora to allow for any statistical evaluation, but the limited available data suggests that the because X construction is informal insofar as it seems to be primarily associated with spoken language (which again might be linked to another pragmatic factor discussed below). However, there is no indication that it is also associated with non-standard, lowstyle language (like forms such as ain’t or multiple negatives would be). Because X is used in spoken language both in informal and in more formal contexts, such as TV or newspaper interviews. The internet meme because science, associated with an academic topic, also supports this idea. In contrast, causal because is not associated with any particular medium; it occurs both in written and spoke English, in formal as well as informal contexts. Because, of course, has informal variants like ’coz, but this again is a purely phonetic/morphological shortening and does not imply any morphosyntactic changes to the causal because construction as such. The second pragmatic difference lies in the use of because X. What is the contrast, if any, between (23a) and (23b)? (23)

a. Lacking that safe spot, conversations with Clarence could be scary, because unscripted. But that wasn’t the absolute worst of it. (COCA 1998, Harper’s Magazine) b. Lacking that safe spot, conversations with Clarence could be scary, because they were unscripted. But that wasn’t the absolute worst of it. (my modification)

I suggest that there is a difference in subjectivity between the two constructions. Subjectivity is a particular pragmatic perspective on a given situation. In the sense of Traugott & Dasher (2004: 97–99), it is the explicit inclusion, by linguistic means, of the speaker’s commitment or epistemic stance towards what is being said. Keller (1995) discusses the notion of subjectivity on the basis of the variability in word order in German sentences with weil ‘because’, illustrated below. In (24a) we find regular, standard word order in the subordinate clause (the finite verb is clause final), whereas in (24b) we find main clause (subject-finite verb) order in the subordinate clause. (24)

a. Er ist nach Hause gegangen, weil er Kopfweh hatte. He is to home gone , because he headache had ‘He went home because he had a headache’ b. Er ist nach Hause gegangen, weil er hatte Kopfweh. He is to home gone , because he had headache ‘He went home because he had a headache’

(Keller 1995: 23)

Keller suggests that the two structures are not synonymous, but that they signify different pragmatic perspectives. Standard word order, as in (24a), represents an

Because science! Notes on a variable conjunction 

 55

objective causal relationship between the two clauses. It focuses on facts: the headache was the cause for his going home. In (24b), with a so-called epistemic “weil”, the fact is turned into an argument. It focuses on the speaker’s knowledge. Factual “weil” in (24a) “gives the reason for his coming home, while [epistemic weil, ex. (24b)] gives the reason for my concluding that he has gone home” (Keller 1995: 22). In this sense, epistemic weil can be said to be more subjective, as it includes the speaker and his/her perspective and evaluation of the situation in the utterance itself. I would suggest that a similar phenomenon, albeit perhaps less strongly, can be seen in causal because versus because X. Consider the constructed utterances (25a) and (25b), in analogy to Keller’s examples. (25)

a. He went home because of a headache. b. He went home because headache.

Are (25a) and (25b) synonymous, or do they offer different perspectives? Is one perhaps more subjective than the other? One of the tests devised by Keller to check whether utterances like (24a,b) or (25a,b) are synonymous involves negation, as in (26a,b). (26)

a. He didn’t go home because of a headache. b. He didn’t go home because headache.

Two alternative readings seem to be available for (26a): “he went home, and he had a headache, but there is no causal link between the two events” or “he did not go home and he had a headache, and the headache is why he did not go home”. In (26b), I would argue along with Keller (1995: 21), only one of the two readings is easily available, namely that the headache was the cause for him not to go home. The first reading (“he went home, and he had a headache, but the two are unrelated”) is not readily available for (26b). This can also be seen in example (3) above, repeated here as (27). (27)

But motivation alone does not assure success “Because circumstances.” (COCA 1996, International gifted women: Developing a critical human resource)

Here it can also be intuited that the negative scope is on the whole sentence rather than the because phrase. Hence, circumstances can only be seen as the reason why motivation does not assure success.9

9 I am very grateful to Elizabeth Traugott for pointing this out to me.

56 

 Alexander Bergs

This becomes even clearer if we use the more complex structure “It is not the case that” introduced by Keller (1995: 20–21). (28)

a. It is not the case that he went home because of a headache. b. It is not the case that he went home because headache.

(28a) again offers both readings, while (28b) more strongly suggests that he did not go home and that this was due to a headache. In other words, it seems that the because X construction encodes greater speaker subjectivity than the corresponding causal because construction. It shows greater focus on speakers and their epistemic stance towards the situation described. We can thus conclude that utterances with causal because and because X are not synonymous, and that there is indeed a difference, at least in the pragmatics, if not also in the semantics of the two constructions. In line with Keller (1995: 23), I posit that apart from this difference in pragmatic meaning, the two utterances in (25a,b) also have different communicative functions. (25a), with causal because, offers the reason for going home. In [25b], however, the speaker “offers his argument for the event stated in the antecedent, as if saying, ‘He has gone home, and I know why, and am going to tell you: he had a headache’” (Keller 1995: 23). The because X construction thus presents the epistemic stance of the speaker more clearly than the corresponding causal because construction and can be seen as more subjective. This also ties in nicely with the fact that because X is more common in spoken and computer-mediated communication. Subjectivity is not strictly restricted to spoken language, but many expressions of subjectivity are typically found in spoken, informal language: inversions are just one such example (Stein 1995), as are discourse markers of the “I think” or “I guess” kind, and degree modifiers like “kinda” and “sorta” (Traugott 1995). De Cock (2015) also points out that spoken language is particularly interesting for the study of (inter-) subjectivity. However, it should also be noted that the jury still seems to be out on this one, and that it is not entirely clear yet if there is such a relationship between spoken language and subjectivity, and of what kind it might be. Spooren et al. (2010), for example, in their study of the two Dutch causal connectives omdat and want, come to the conclusion that these two connectives differ in their degree of subjectivity, but that this is not in any way related to the medium (spoken versus written). They assume, rather, that the degree of planning lies at the heart of the matter. However, since the relationship between spoken and written language, planned and unplanned discourse, conceptually oral versus conceptually written language, is a particularly tricky one and needs extensive clarification, this question cannot be discussed at this point. Yet this issue certainly warrants investigation in the future.

Because science! Notes on a variable conjunction 

 57

Because conclusion In this paper, I have shown that because indeed occurs in a number of different constructions, ranging from the regular, standard causal because through various micro-constructions of the because N/A/particle kind, to a meso-construction because X. It has also been shown that these are not necessarily new constructions in twentieth-century English, and that some of the micro-constructions (because A in particular) can be traced back at least to the nineteenth century. However, Present-day English seems to have seen a spread of the micro-construction(s) and the development of the meso-construction, partly because of increased type frequency, but also because of prominent and popular singular constructs such as because science or because reasons that serve as attractors in a kind of snowcloning effect. It has also been argued that causal because and because X are related but different constructions which may be partly synonymous on the propositional level, but which are different on the pragmatic level. While causal because is a more factual, objective construction, because X is more subjective and expresses epistemic stance on the part of the speaker. Unfortunately, even though the constructions seem to be quite well-known and popular (especially among younger speakers and in computer-mediated communities) we still do not have enough reliable data from spoken (and written) language. Once more data becomes available, future research will have to look again into questions such as subjectivity, orality, economy and language planning as possible factors in the development of the construction. Also, we find the well-known fragment because/’coz Ø construction (Kid: ‘Why can’t I have another marshmallow?’ Parent: ‘Because!’) which is used when a speaker is unable or unwilling to give particular reasons for a fact or a decision. I suspect some complex relation to the because X construction discussed here, but this will have to be the topic of a future paper. In any case, both because/’coz Ø and because X seem to suggest that the venerable concept of subordination (based mostly, I suspect, on traditional ideas of Latin grammar) might need some re-evaluation (see introduction, this volume). While we still see regular and predictable subordination patterns in (formal) writing, real language in spoken interaction offers us some challenges by presenting structures which could be aligned on a gradient from traditional subordination (causal because) through apposition (because X) to fragmental insubordination (because/’coz Ø). Prosody and intonation might be relevant factors here, but have been ignored entirely in the present paper, mostly due to a lack of data. Intuitively, one might suspect that because X may have a slightly longer pause before X than the corresponding causal because construction, and that because/’coz Ø carries a particular intonation. But, once more, this will have to remain a question for further research, because science.

58 

 Alexander Bergs

References Bailey, Laura R. 2014. ‘Because x’: Syntactic restructuring, ellipsis, or internetese? Paper presented at the Annual Meeting of the Linguistics Association of Great Britain, Oxford, 4 September. Bailey, Laura R. 2012. Because reasons. linguistlaura. https://tinyurl.com/p37uwjw (accessed 26 April 2016). Bergs, Alexander & Thomas Hoffmann. 2014. Are you a construction in disguise? – Soziale und physische Kontexteigenschaften von Fußballgesangskonstruktionen. In Alexander Ziem & Alexander Lasch, (eds.), Konstruktionsgrammatik IV: Konstruktionen als soziale Konventionen und kognitive Routinen, 115–131. Tübingen: Stauffenburg. Bergs, Alexander & Thomas Hoffmann. 2018/in print. A Construction Grammar approach to genre. Cognitextes Bohmann, Axel. 2016. Language change because Twitter? Factors motivating innovative uses of because across the English-speaking Twittersphere. In Lauren Squires (ed.), English in Computer-Mediated Communication: Variation, Representation, and Change, 149–178. Berlin/Boston: de Gruyter Mouton. Bolinger, Dwight. 1968. Aspects of Language. New York: Harcourt Brace. De Cock, Barbara. 2015. Subjectivity, intersubjectivity and non-subjectivity across spoken language genres. In Bert Cornillie & Barbara de Cock (eds.), Hearer-orientation in spoken genres. Special Issue of Spanish in Context 12 (1). 10–34. Goldberg, Adele. 1995. Constructions. Chicago: University of Chicago Press. Goldberg, Adele. 2006. Constructions at Work. Oxford: Oxford University Press. Goldberg, Adele & Ray Jackendoff. 2004. The English Resultative as a Family of Constructions. Language 80. 532–568. Hirose, Yukio. 1991. On a Certain Nominal Use of Because-Clauses: Just Because BecauseClauses Can Substitute for That-Clauses Does not Mean that This is Always Possible. English Linguistics 8. 16–33. Hoffmann, Thomas & Graeme Trousdale (eds.). 2016. The Oxford Handbook of Construction Grammar. Oxford: Oxford University Press. Hoffmann, Thomas. 2017. Construction Grammars. In Barbara Dancygier (ed.), The Cambridge Handbook of Cognitive Linguistics, 310–329. Cambridge: Cambridge University Press. Kanetani, Masaru. 2015. On the New Usage of Because. Studies in Language and Literature [Language] 68. 63–80. Keller, Rudi. 1995. The epistemic weil. In Dieter Stein & Susan Wright (eds.), Subjectivity and Subjectivisation: Linguistic perspectives, 16–30. Cambridge: Cambridge University Press. Plag, Ingo. 2006. Productivity. In Bas Aarts & April McMahon (eds.), Handbook of English Linguistics, 537–556. Oxford: Blackwell. Pullum, Geoffrey. 2014. Because syntax. Language Log. http://languagelog.ldc.upenn.edu/ nll/?p=9494 (accessed 26 April 2016). Pullum, Geoffrey & Rodney Huddleston. 2002. Prepositions and Preposition Phrases. In Rodney Huddleston & Geoffrey Pullum (eds.), The Cambridge Grammar of the English Language, 597–663. Cambridge: Cambridge University Press. Quirk, Randolph, Sidney Greenbaum, Svartvik, Jan & Geoffrey Leech. 1985. A Comprehensive Grammar of the English Language. London: Longman. Schnoebelen, Tyler. 2014. Innovating because innovation. https://corplinguistics.wordpress. com/2014/01/15/innovating-because-innovation/ (accessed 13 February 2017).

Because science! Notes on a variable conjunction 

 59

Spooren, Wilbur, Ted Sanders, Mike Huiskes & Liesbeth Degand. 2010. Subjectivity and Causality: A Corpus Study of Spoken Language. In Sally Rice & John Newman (eds.), Empirical and Experimental Methods in Cognitive/Functional Research, 241–255. Stanford: CSLI Publications. Stein, Dieter. 1995. Subjective meanings and the history of inversions in English. In Dieter Stein & Susan Wright (eds.), Subjectivity and Subjectivisation: Linguistic perspectives, 129–150. Cambridge: Cambridge University Press. Sweetser, Eve. 1990. From etymology to pragmatics. Metaphorical and cultural aspects of semantic change. Cambridge: Cambridge University Press. Traugott, Elizabeth Closs. 1995. Subjectification in Grammaticalization. In Dieter Stein & Susan Wright (eds.), Subjectivity and Subjectivisation: Linguistic perspectives, 31–54. Cambridge: Cambridge University Press. Traugott, Elizabeth Closs & Richard B. Dasher. 2004. Regularity in Semantic Change. Cambridge: Cambridge University Press.

David Denison

That-clauses as complements of verbs or nouns Abstract: Many types of verb in English permit a that-clause complement, including factual and suasive verbs like acknowledge, point out, recommend, suggest, etc. (1) I assume she acknowledges that it was a blunder (2015, COCA) I will refer to this construction as ‘V + that’. Other transitive verbs like advance, contest, contradict, endorse, highlight, moot, pose, put forward, propound and uncover are not supposed to take thatclauses, and by and large in the pre-2000 corpora they don’t. Instead such verbs take a nominal object, one form of which can be a “shell noun” (abstracts like fact, claim, argument, situation) with a that-clause as complement of the noun, a construction referred to here as ‘shell + that’, as in (2). (2) it highlighted the fact that this is a unique case (1995, COCA) In recent English, simple that-clause complements are spreading to at least some such verbs in the active, as in (3). (3) a. In the first issue of Salvage, Neil Davidson mooted that neoliberalism may be undermining the basis for capital accumulation itself. (2015, China Miéville, ‘On Social Sadism’, Salvage http://salvage.zone/ in-print/on-social-sadism/ 17 Dec 2015) b. The renowned Millennium Ecosystem Assessment […] highlights that coasts are the most highly degraded ecosystems on our planet. (2005, COHA) In this paper I explore data from COHA concerning the development from shell + that to V + that, discussing the difficulties of getting reliable evidence here. The topic also raises a tricky problem for corpus linguists, namely whether it is ever legitimate to talk of native speakers not being in full control of some registers of their own language.

David Denison, University of Manchester https://doi.org/10.1515/9783110583571-004

62 

 David Denison

1 What? I begin by explaining what usage is to be discussed (Section 1), move on to how it might have come about (Section 2), detail my corpus investigation into when it appeared (Section 3), then raise questions about whose grammar is involved and what that might mean (Section 4), before a brief consideration of what else needs doing (Section 5). Many factual and suasive verbs and verbs of communication can include a finite-clause proposition in their complement, either directly as a that-clause, (1), or indirectly with a that-clause dependent on an object noun, as in (2):1 (1)

claim that X, say that X, suggest that X

(2) advance the claim that X, formulate the suggestion that X, welcome the fact that X Such object nouns have been defined in various ways and given a number of different labels, most recently “signalling nouns” (Flowerdew & Forest 2015), though I will stick to the more widely used “shell nouns” (Hunston & Francis 2000: 185–6, Schmid 2000).2 My concern here is with the spread to certain new verbs of pattern (1) complementation (henceforth ‘V + that’) at the expense of (2) (‘shell + that’). I present some evidence of recent change and provide preliminary answers to the questions where, when, how and why. For this I conduct a systematic investigation in the Corpus of Historical American English (COHA, Davies 2010–). According to Biber et al. (1999: 668), “[…] communication verbs controlling that-clauses (apart from say) are most frequent in academic prose”, and indeed my first exposure to the usage which is the topic of this paper was in essays by British students. Many of the writers who produced the examples in question were relatively unskilled, insecure about written expression, and they may have been using a word-processor thesaurus for “elegant variation” or to avoid the risk of 1 There are, of course, introducers of finite clause complements other than that, notably whether and how. 2 Hunston & Francis (2000: 186) characterize shell nouns semantically as follows:   The idea that they are a distinct and definable group – arguably a class of their own – is reinforced by the discovery that all these nouns have features of meaning in common. They fall into two major groups. The first group consists of nouns which refer to something that is written or spoken; the that-clause lexicalises what it is that is written or spoken. The second group consists of nouns which refer to beliefs, ideas, wishes, and thought processes. Again, the that-clause lexicalises what it is that is thought or believed.

That-clauses as complements of verbs or nouns 

 63

plagiarism. Here are some genuine examples which contain a verb used directly with a that-clause complement which to my ears needed an NP complement. (The names of cited scholars are replaced by X to avoid distraction.) (3) X et al. (1985) advance that the try in try and V is […] (2009) (4) X (1970) notes how this order of events is valid, citing that […] (2015) (5) Along with X who contested that people who reached for […] (2012) (6) […] with X criticising that the French influence was sporadic […] (2010) (7) X defines that “grammar may be regarded either from a theoretical or practical point of view. […]” (2011) (8) Within this discussion he deliberates that “‘Correctness’ as it is popularly conceived, is one of these terms that seems to refer to something desirable, […]”. (2011) (9) The OED entry depicts that the word was originally borrowed from French […] (2014) (10) X expressed that Middle English was indeed a creole […] (2010) (11)

John Cheke who was a professor of Greek at the University of Cambridge instigated that our language should be written cleane & pure’. (2012) [punctuation as shown; intended meaning of verb unclear – DD]

(12)

This portrays that the preacher deserves pity. (2011)

(13)

X poses that for males the same interruptions may act as a bid for the floor (2013)

(14)

[…] which can be reinforced by X et al, who utters that, “In other locations […]”(2010)

Since I gave presentations on this phenomenon in Denison (2009, 2011), I have collected further examples, often from journalism, to expand the range of verbs involved. Nevertheless, that collection is unsystematic and limited to a couple of examples of each verb. As for the supposed standard Present-day English (PDE)

64 

 David Denison

Table 1: Complementation by shell + that and/or by that-clause alone. Verb

shell + that

accept, acknowledge, add, affirm, allege, allow, announce, assert, assume, believe, concede, confirm, consider, convey, demonstrate, deny, disclose, discover, doubt, emphasise, establish, explain, forget, guarantee, hold, imagine, infer, maintain, mention, observe, ?pronounce, propose, prove, recognise, regret, repeat, report, see, state, stress, submit, suggest, suspect, understand claim, deduce, determine, find, indicate, judge, point out, predict, presume, show, stipulate, suppose advance, articulate, back up, challenge, communicate, contradict, convey, define, discuss, dispute, encourage, endorse, enlarge upon, espouse, express, oppose, promote, put across, put forward, question, rule, support, sustain, underline, underscore, utter advocate, analyse, bring to the surface, cite, clarify, contest, criticise, deem, deliberate, depict, describe, display, exemplify, explicate, highlight, identify, illustrate, inform, instigate, interpret, moot, portray, pose, posit, propound, publicise, quote, reflect, refute, reinforce, reiterate, respect, rule out, solidify, stand, summarise, take into account, uncover, update, view, welcome, yield

? (≠)

V + that

(≠) ×

?

?

against which I was comparing this usage, I tried to gauge the scholarly consensus by combining relevant lists of verbs in Biber et al. (1999: 362–3, 660–70), Huddleston & Pullum (2002: 958–9, 1017–22), Quirk et al. (1985: 1176–7, 1180–83) and Collins COBUILD (1987).3 I also conducted searches in the OED and a number of corpora for a few specific verbs. Verb subcategorization in English is partly systematic but in part apparently arbitrary. Huddleston & Pullum, for example, devote three chapters to different patterns of complementation (2002: Chapters 4, 11, 14), offering a combination of explanatory principles and mere lists. The other major grammars (Quirk et al. 1985, Biber et al. 1999) have comparable expositions. Dictionaries organize similar material by verb rather than by complementation pattern. Possible complementation patterns are a particular focus of learner dictionaries. My tabulation of information in the handbooks is as shown in Table 1. The first group of verbs are reported as allowing either shell + that or V + that, 3 Note that report-cl in COBUILD covers more than just that-clauses.

That-clauses as complements of verbs or nouns 

 65

while the second group have limited occurrence in shell + that, not synonymous with use in V + that. The third group is reported as allowing shell + that but not V + that (though dispute and question can occur with a whether-clause). I have marked in bold those that I have attested with a that-clause – contra the handbooks. The relevant behavior of the fourth group is not specifically mentioned in the handbooks (apart from inform occurring with an indirect object and that-clause) and therefore is in particular need of corpus study. These are verbs which I have attested in V + that but which in 2011 I felt would not be possible in my idiolect. The unreliability of introspection is such that I would no longer reject with any certainty deem, posit, reiterate or take into account.

2 How? The verbs that can occur in patterns (1) and (2) are by no means mutually exclusive sets. Many verbs (listed in the first row of Table 1) take part in both patterns with very similar meanings either way. (15)

a. accept that X, disclose that X, prove that X, understand that X b. accept the claim that X, disclose the fact that X, prove the case that X, understand the observation that X

This point is clearly crucial to any explanation of the spread of the V + that construction. The simplest mechanism responsible would be analogy: the dual subcategorization of verbs like those in (15) is extended to verbs like those in (2). If analogy is a mechanism at work here, it raises the question of timing. Why should an analogy which has existed for centuries prove so productive only now? Has the spread merely become more noticeable because we now have access to more unedited material from less experienced writers, or is it that less experienced writers are taking advantage of online thesauri which offer a variety of verb synonyms but are silent on their complementation patterns?4 Then again, is the development in question actually as recent as it seems? Perhaps that impression

4 A resource like the Oxford Advanced Learner’s Dictionary would give usage guidance on a verb, while the Language Bank’s website could suggest synonyms, but I don’t know how long such services have been in existence online. In any case I doubt that many British native speakers would think of using a resource explicitly aimed at language learners; see Section 4.

66 

 David Denison

was mistaken, and there is a longer-term tendency for verbs to add the possibility of that-clause complementation.5 Zwicky (2005) has two salutary warnings here: […] the Recency Illusion, the belief that things YOU have noticed only recently are in fact recent. This is a selective attention effect. Your impressions are simply not to be trusted; you have to check the facts. Again and again -- retro not, double is, speaker-oriented hopefully, split infinitives, etc. -- the phenomena turn out to have been around, with some frequency, for very much longer than you think. It’s not just Kids These Days. Professional linguists can be as subject to the Recency Illusion as anyone else. […] Another selective attention effect, which tends to accompany the Recency Illusion, is the Frequency Illusion: once you’ve noticed a phenomenon, you think it happens a whole lot, even “all the time”. Your estimates of frequency are likely to be skewed by your noticing nearly every occurrence that comes past you. People who are reflective about language -- professional linguists, people who set themselves up as authorities on language, and ordinary people who are simply interested in language -- are especially prone to the Frequency Illusion.

Note in addition that the simple analogical explanation would not cover deliberate that X in (8), since deliberate is an intransitive verb which does not occur in shell + that, while the sense of instigate in (11) is so obscure that the example should probably be rejected outright as an error. However, the whole question of grammaticality vs. error is central to this investigation and cannot be taken for granted; I take this point up in Section 4. Returning to the question of how new V + that usages might have arisen, the simple analogy mentioned above is not the only possible mechanism for some verbs. Newmeyer (2003: 160 n.12) is one scholar who has pointed out that A number of verbs disallow sentential complementation in surface direct object position, but do allow them as subject of passives: (i) (ii) (iii)

a. b. a. b. a. b.

*The grammar expresses [CP that the rule is obligatory]. [CP That the rule is obligatory] is expressed by the grammar. *This formulation of the rule reflects [CP that all NPs behave uniformly] [CP That all NPs behave uniformly] is reflected by this formulation of the rule. *This theory captures [CP that languages are learnable]. [CP That languages are learnable] is captured by this theory.

5 This possibility was underscored by two examples from the Sherlock Holmes stories of Conan Doyle provided by Professor Minoji Akimoto after my lecture in Tokyo (Nov. 2016): (i) I even distinguished that one of them was marked with the name of ‘Hyams’, who was Oldacre’s tailor. (1903, The Adventure of the Norwood Builder) (ii) And in this way he managed that your good man should have no want of drink [...] (1892, The Adventure of the Copper Beeches)

That-clauses as complements of verbs or nouns 

 67

Newmeyer’s examples have a that-clause as passive subject, a structure which is rather high-flown. Extraposed passive examples with a dummy it-subject are less rare, (16), and passive participles also occur in the complement of other verbs, as in (17); see also (22)–(24), (29) below. (16) but it must be faced that I have doused all hope of continued revelry (1951, COHA) (17) Don’t you want it known that you’re all right? (1921, COHA) In the light of these facts, a conceivable route to some new cases of V + that would be for a subcategorization already licensed for a passive matrix verb to be extended to the same verb in the active. Although the investigations described in Section 3 threw up a modest number of passive examples, they were not numerous enough for their relative frequency to be tested statistically, so their possible relevance to the entry and spread of new V + that patterns remains unknown. Intuitively, though, what are often rather elaborate, formal-sounding passives are rather foreign to the style at least of the student writers we have mentioned, making it unlikely that the passive would have been an important conduit for their usage at least. For some of the verbs where V + that is unexpected, Erik Smitterberg (at LMEC6, 2017) made a valuable comparison between a that-clause and a finite clause introduced by how, asking speculatively whether the existence of V + how could somehow have helped to license V + that. I give parallel examples in (18) and (19). (18)

a. I love how they’re always using the phone. (1970, COHA) b. but he couldn’t stand how she refused to take her own life seriously (1989, COHA)

(19)

a. I love that she kept some of the funny wallpaper in her house (2004, COHA) b. and then you couldn’t stand that I was becoming successful (2002, COHA)

While most how subclauses have a manner or degree reading in line with the usual meanings of how, there are some – such as those I have cited in (18) – where the meaning of how approaches the purely factitive meaning of that. A possible route to the spread of V + that would then be as follows: manner how gets bleached to a neutral subordinator and is then replaced by the more general subordinator that.

68 

 David Denison

We might go further and compare the way (that) (see introduction, this volume), which is semantically similar to how in allowing in suitable contexts both a manner reading (‘the manner in which’) and a bleached, factitive meaning (‘the fact that’). (20)

and I resented the way that accidents had snatched their dignity from them (2006, COHA)

Then we would have a route from shell + that, as in (20), to V + how, as in (18), to V + that.6 I have not explored scenarios involving how in any detail, but likely problems include relative dates and low numbers: I suspect that some supposed precursors are no older than the V + that patterns to be explained, and it looks as if types (verbs which can subcategorize for factitive how) and tokens are both infrequent. For this paper I concentrate on the direct analogical route from shell + that to V + that.

3 When? This section concerns the timing and frequency of the linguistic development, which inevitably involves a detailed description of the methodology employed.

3.1 Problems with an online search To see how reliable the handbooks are in their characterization of what is and isn’t possible, we need a baseline survey of PDE corpora. And we need to survey historical corpora to document the advent and rise of that-clause complementation with verbs that had not previously permitted it. String searches would be too inefficient: many of the verbs in question are identical in form to nouns, and the word that is too multifunctional. I therefore start with a large tagged historical corpus, COHA. As far as I am aware, this is currently the only corpus with historical coverage up to the present which is large enough to provide a sufficiently wide range of data.

6 A referee points out a more direct route, namely reduction of the way that to simple that.

That-clauses as complements of verbs or nouns 

 69

A search for [display].[v*] that.[cst] in COHA specifies the verb display in any of its forms and the conjunction that (which includes the relativizer function). Of 41 hits, 18 have the noun display wrongly tagged as a verb,7 and another 18 have determiner that. The accuracy of tagging in COHA clearly leaves something to be desired, and here we are looking for a construction which may be innovative for the verb display and absent from the tagger’s training corpus, but an accuracy of 5/41 or 12% is too low to be helpful. Furthermore, of those five that-clauses following the verb display, one is not in construction with that verb, and three are relative clauses, so in the end only one is at all relevant. (21)

The man made the animal open its mouth as wide as possible to appear fearsome, but all this accomplished was to display that the poor creature’s teeth have been yanked out. (1990, COHA)

A similar search for [advance].[v*] that.[cst] produces 137 hits, of which 56 have the probable adjectival participle advanced, 13 have the noun advance, 2 have determiner that. Of the 66/137 correctly tagged (48%), 58 are irrelevant, leaving eight cases for consideration. Of these, three involve a passive with an extraposed that-clause. (22)

We hav[e] nowhere seen it advanced that even a minority of the Legislature of 1839 made any movement of Repudiation which might at least have served as a notice of intention. (1842, COHA)

(23)

Since Hannah’s death I have heard it openly advanced that she was the guilty party in the crime (1878, COHA)

(24)

It has been advanced that great queens owed their power to the association and advice of the noble and high-minded men who surrounded them (1911, COHA)

And five examples, all early, do seem to have the verb advance in the active with a that-clause complement. (25)

Buffon advanced that the brain of the orang-outang “does not differ from that of man (1835, COHA) [writer of this and two other examples is called Franz Josef Gall]

7 However, a search for [display].[n*] that (the noun display followed by any that) produces 44 hits, while [advance].[n*] that gives326, and every single example indeed has a noun and not a mistagged verb. This will be important in §3.3; see note 9.

70 

 David Denison

(26)

If some of them /q/ have advanced that every thing proceeded from corruption, others, more honest and sincere, have refuted them, even in the earliest times. (1836, COHA)

(27)

Some have advanced that the finest roses from seed are always the longest in flowering (1844, COHA)

Finally here, in the hope of higher precision I ran a similar search for a verb, define, which does not have a noun homonym. There are 42 hits for [define].[v*] that.[cst], of which six have the probable or possible adjectival participle defined, three have pronominal that, nine have determiner that, and one is removed by the software as a probable duplicate. Of the 23/42 remaining (55%), two have relativizer that and 15 are irrelevant, leaving four early examples of define + that-clause, (28), one with passive define and extraposed that-clause subject, (29), and one fairly recent occurrence, (30). (28)

The Pope defined that it was a revealed truth that Mary had been conceived without incurring original sin (1861, COHA + 3 other examples in the same text)

(29)

It is defined that Almighty God is auctor utriusque Testamenti. (1882, COHA)

(30)

How to define that it has a kind of off-reddish tint, neither quite one color nor another, stumps me until […] (1993, COHA)

From these brief experiments with tagging we see that precision in COHA searches for this kind of material can be poor – especially when the construction might be a recent innovation. To gauge recall, we would have to repeat the test for the lexemes display/advance/define with tags other than verb or with that tagged as something other than conjunction. In effect, we would have to discard the tagging. Given the difficulty of conducting an efficient search in online COHA or a sister corpus like GBooks, I have considered recourse to parsed corpora. Possibilities include (i) the Penn Parsed Corpus of Modern British English, 2nd edition, release 1 (PPCMBE2), for the period up to 1914, and (ii) the Diachronic Parsed Corpus of Spoken English (DCPSE), to compare the 1960s and 1990s, in both of which the part of speech (POS) tags are manually verified. There is also (iii) the dependency-parsed version of ARCHER 3.2 at Zurich, for the period 1600–1999. However, because of their dates and modest sizes, none of these parsed corpora could offer any evidence relevant to the suspicion that online thesauri play a role

That-clauses as complements of verbs or nouns 

 71

in the spread of V + that, so I am thrown back on COHA for its size and its perfect chronological range, but necessarily the offline version.

3.2 Identifying verbs in V + that We need a systematic, corpus-based means of identifying verbs that appear in shell + that and which might also allow V + that. Rather than collecting every possible that-clause complementing a verb, which would have been an overwhelming task, I started instead from the presumed source of the innovation, pattern (2), shell + that. I also limited my research to the twentieth century plus the first decade of the twenty-first, thus the last 11 decades of COHA. In Flowerdew & Forest (2015) there are convenient lists of the shell nouns found in an academic corpus of PDE. I noted the 50 most common of these.8 The procedure I eventually settled on was to search the largest and most recent decade of COHA, 2000–2009, for shell + that involving those shell nouns. The search, programmed in Python, strips out most punctuation and looks for suitable strings, with additional words of context to left and right. In effect – to translate the actual Python formulation into more familiar terms – the crucial portion looks for the following sequence in the corpus. (31)

verb + article + shell-noun + that …

The program only uses POS tags to pick out “verb”9 and “article” in (31) and to ensure that the verb is not be; the slot “shell-noun” draws in a list of items, and the unreliable POS tagging of that is ignored. On the first run I used the 20 most frequent shell nouns. The resulting hits were relatively quick to filter, leaving 604 (64% precision) that did exemplify shell + that. I sorted these by matrix verb to find the most common verbs introducing shell + that. Figure 1 shows those verbs occurring five times or more.

8 The 50 most common shell nouns listed in Flowerdew & Forest (2015: 86–7, Table 8.2), in descending order of frequency, are case, way, problem, result, thing, theory, model, idea, example, effect, point, reason, equation, question, process, issue, method, right, fact, principle, condition, approach, argument, procedure, change, difference, strategy, situation, solution, time, policy, analysis, view, decision, evidence, possibility, role, function, area, expectation, concept, ability, factor, conclusion, need, consequence, section, stage, failure and response. I added the alternative spelling rôle. 9 From my experiment with display (note 7 above), it seemed unlikely that there would be many nouns mistagged as verbs. The reverse error is common.

72 

 David Denison

At first I investigated those verbs with six or more occurrences (left of the vertical line in Figure 1).10 When I came to see how often they occurred in V + that over the period 1900–2009, the results were disappointing: 1,745 good examples (precision 73%), with individual totals very low for all but four verbs; indeed three out of 20 had none at all, while six more occurred ten times or less in V + that. Nor was there much in the way of chronological patterning. Selecting verbs on the basis of probable frequency in shell + that had not been particularly enlightening. 80 70

N = 604

60 50 40 30 20 10

have ignore like accept make get hide change support face consider appreciate hate overlook reflect lament reinforce alter disguise love mention miss resent absorb address bemoan confirm discuss forget grasp regret respect underscore

0

Figure 1: Most common verbs with shell + that 2000–2009.

To make the list of suitable verbs more comprehensive, I added all remaining verbs with between 1 and 5 genuine examples of shell + that from the initial shell noun search: 136 verbs,11 of which 51 would actually turn out to appear in V + that 1900–2009. I also ran the shell noun search again with nouns 21–50 from the list in footnote 10, once again only for the 2000s decade, but this time quickly sifting the  hits for just one valid example of shell + that with any matrix verb not yet included in the research. This added a further 36 verbs, of which 21 later appeared 10 The 20 verbs are accept, alter, appreciate, change, consider, disguise, face, hate, hide, ignore, lament, like, love, mention, miss, overlook, reflect, reinforce, resent and support. The common get, have (generally occurring with shell idea) and make (with case or point) were excluded as not offering any synonymous V + that possibilities. 11 The list actually fed to the program includes a few extra verb lemmas to allow for spelling variants like criticise/criticize, savor/savour, as COHA is inconsistent about lemmatizing such pairs together.

That-clauses as complements of verbs or nouns 

 73

in V + that, giving a grand total of 89 verbs which had occurred in COHA in shell+ that at least once in the decade 2000–2009. Finally, from my own collection of attested examples with “unexpected” (to me) V + that rather than the normal shell + that, I added 15 more verbs, 9 of which appeared in the V + that searches of COHA.

3.3 Counting V + that The search procedure for V + that was broadly similar to that for shell + that, except that this time it proved necessary to increase the initial precision (thus lowering the recall) by restricting what follows that to some of the more probable complement that-clause shapes:12 (32)

a. b. c. d. e. f.

matrix-verb + that + pronoun + V … matrix-verb + that + existential-there + V … matrix-verb + that + N + V … matrix-verb + that + N + N + V … matrix-verb + that + article/possessive + N + V … matrix-verb + that + determiner + N + V …

The program is fed a list of items for the “matrix-verb” slot; the word that is just a string; and the remaining items use the POS-tags of COHA. The rough-and-ready characterization in (32) of likely that-clause openings successfully cuts out many examples which are not V + that, though of course at the cost of some good examples, e.g. those where the subject NP of the that-clause includes a post-modifier, or where an adverbial intervenes after that. Compare an online search like [respect].[v*] that, where the verb is tagged but that is not. From a few experiments with different verbs, it looks as if such an online search typically retrieves about 2.5 times more valid V + that examples than the Python program just described. With an online string search like respect| respects|respected|respected|respecting that, the haul of V + that might possibly be slightly greater still, but that would have to be set against the fact that with this verb, every single form has a non-verb homonym. Use of the verb tags is certainly worth it. Both offline and online searches make use of tagging of the matrix verb, so both would miss an example if the verb had been mistagged as a noun. In practice, neither the COHA server nor I have the patience for what would 12 Jiří Zámečník’s programming makes ingenious use of CLAWS tags to avoid the processing in Python of lengthy regex formulas. A similar restriction was tried out in the shell noun search but was found to lower recall more than it increased precision.

74 

 David Denison

have been over 170 online searches, and once there are many thousands of hits to deal with, precision becomes crucial. For verbs that are frequent in V + that, the Python search summarized in (32) shows almost 100% precision and was the only practical choice. My assumption is that the sample of V + that obtained will nevertheless give a fair picture of chronological trends. So far – and indeed throughout this paper – I have only allowed for cases where the that-clause follows on immediately from the shell noun in shell + that or the matrix verb in V + that. With V + that, therefore, my searches will not pick up examples like (33)

a. That they are fair I can testify (1900, COHA) b. That the argument is sound can be seen by exhibiting its structure (1973, COHA)

(34)

a. You mentioned earlier that he’d summoned you. (1996, COHA) b. but he remembered with a slight wince that it had cost $1,800 (1953, COHA)

I will simply ignore pre-placement of that-clauses as in (33), on the assumption that any small increase in recall – for they are not common – would be vastly outweighed by a loss of precision; that is, if the search could be programmed at all. What about that-clauses that follow their head but with something interrupting, as in (34)? I tested this for both early stages of my procedure. I ran a string search in online COHA for the noun case followed within nine words by that and took a sample of 200 from the 13,015 hits. The sample contained 11 examples of shell + that, in ten of which that was adjacent to case, as against just one clear example where it was non-adjacent.13 Similarly, I searched for the verb lemma accept followed within up to nine words by that, this time in the period 1900– 2009, and took a random sample of 200 from the 5,402 hits: 25 out of 200 exemplified V + that, and in 23 of those 25, V and that were adjacent. It does seem reasonable, then, to limit searches to adjacent that. One danger of confining searches to that-complements that immediately follow their head is that syntactic change is often thought to begin “sneakily”, in contexts where a change is least noticeable (De Smet 2012: 605). If such is the case 13 I discounted a second, doubtful non-adjacent example, which probably involves complementizer so that:   (i) I say then, that, even though the case could be so that the whole system of Catholicism was recognized and professed, without the direct presence of the Church, still this would not at once make such a University a Catholic Institution […] (1852, COHA)

That-clauses as complements of verbs or nouns 

 75

here, there is a risk of missing some crucial early examples. But statistically, and for reasons of sanity, the simplification of search strategy is justified. All searches for V + that reported here were made with the same Python script as sketched in (32). The program takes a list of verbs and works with one decade of COHA data at a time. Most of the interesting chronological changes involve very low frequencies, but numbers of hits can be very large: one run produced 5,528 for the 1900s alone, with many of the verbs showing hundreds of examples each in that decade: see ×1240, realize/-ise ×427, declare ×422, admit ×288, and so on. These numbers would overwhelm and mask any of the innovatory uses. What it confirms is that V + that is already deeply entrenched by 1900. I decided to split the list of verbs at an arbitrary threshold, but one strongly suggested by the data. Any verb which occurred in V + that more than 20 times in both the 1900s and 1910s was treated as entrenched, and numbers of hits for all of them were combined – between 4,700 and 6,800 per decade – in order to provide a baseline measure of the frequency of V + that over the period. I could only sample the hits for these high-frequency verbs, but after scanning hundreds I had only spotted one non-example, so it seemed safe to count without filtering. For all but a couple of the remaining verbs in the list, there were few or no hits for V + that in the 1900s or 1910s, with most hits that did occur having to be rejected. This is not unexpected for a usage which is recent in PDE and of dubious grammaticality. For these verbs, therefore, precision is unreliable, and all hits over the whole period were filtered for genuine cases of V + that and counted separately.14

3.4 History of V + that 1900–2009 I present the results for different groups of verbs. The 24 “entrenched” verbs (acknowledge, add, admit, assert, consider, declare, demand, explain, fear, find, forget, indicate, learn, mention, propose, protest, prove, realize, recognize, reflect, remember, see, suggest and understand) are shown in Figure 2. The total frequency per million words falls somewhat after the 1920s but remains relatively high by comparison with most of the other verbs in the construction. There was no individual verification of examples.

14 Naturally, decisions have to be made about what constitutes a valid example of V + that. I discounted examples where a participle has developed into a conjunction (‘Given that …’, ‘Providing that …’), and such cases as ‘but it afterward developed that he was never near the place’ and ‘It registers that Amina is no longer in the car’, where dummy it is subject of an active verb.

76 

 David Denison

300 250 200 150 N = 59179 100 50 0

1900s 1910s 1920s 1930s 1940s 1950s 1960s 1970s 1980s 1990s 2000s

Figure 2: V + that entrenched by 1900 (24 verbs, pmw).

Another seven verbs did not meet my criterion for entrenchment in 1900 but became fairly frequent later in the century (accept, concede, disclose, emphasize, establish, repeat and stress); see Figure 3. They too showed a fall, if starting slightly later – after the 1960s or 1970s – apart from accept, which starts very low and rises rapidly after the 1960s. With this group, all examples were verified. The remaining 137 verbs in the list had also produced few or no hits in the 1900s or 1910s, but genuine examples of V + that in later decades were not numerous. With these verbs the precision of the search was poor, and again all hits had to be checked and filtered. The totals per decade are shown in Figure 4.15 The most frequent of the 52 verbs charted are appreciate (72 examples 1900–2009), lament (52), like (52), comprehend (43), advertize (41) and betray (32). Within this group there is a noticeable overall rise in the last four decades. To focus on it I select the 12 verbs from the group that show an individual rise (appreciate, clarify, document, hate, highlight, ignore, like, love, miss, offer, posit and share) plus the one medium-frequency verb that does so, accept (already charted in Figure 3), and plot them in Figure 5, showing raw numbers this time.

15 The list is advertise, appreciate, bear, betray, celebrate, change, comprehend, conceal, contemplate, contradict, corroborate, count, curse, deplore, develop, discount, disguise, dislike, disprove, dispute, disregard, echo, endorse, enjoy, entertain, face, formulate, hate, hide, ignore, include, lament, like, love, miss, offer, omit, overlook, ponder, posit, postulate, press, question, register, resent, sell, share, spot, symbolize, tout, underline and voice.

That-clauses as complements of verbs or nouns 

 77

4.0 3.5 3.0

N = 1807

accept concede

2.5

disclose

2.0

emphasize

1.5

establish repeat

1.0

stress

0.5 0.0

1900s 1910s 1920s 1930s 1940s 1950s 1960s 1970s 1980s 1990s 2000s

Figure 3: Medium-frequency verbs (7 verbs, pmw). 6 5

N = 619

4 3 2 1 0

1900s

1910s

1920s 1930s 1940s

1950s 1960s 1970s 1980s 1990s 2000s

Figure 4: Low-frequency verbs (52 verbs, pmw).

Another ten verbs appear for the first time in V + that in either the 1990s or the 2000s, but with just one example (disprove, disregard, entertain, spot, symbolize, tout, voice) or two examples (change, contradict, dislike) each. Given more data, it might turn out that these verbs too are part of a recent uptick in V + that. I have left them out of Figure 5 to keep it readable. The remaining verbs do not appear to be part of the recent uptick, and I have nothing to say about their individual trajectories. In the light of these figures, what can we say about the history of V + that with our 98 verbs, all of which had appeared in shell + that? Note that the figures only represent instances found with one Python script, perhaps about two fifths of the instances in COHA. The general trend is a gentle decline, although the histories of quite a few verbs suggest a recent uptick in frequency, in some cases even an accretion to the construction. The verbs in Figure 5 (apart from posit and perhaps

78 

 David Denison

80 accept

70 60

N = 423

appreciate clarify document hate

50

highlight

40

ignore

30

like

20

miss

love offer

10 0

posit share 1900s 1910s 1920s 1930s 1940s 1950s 1960s 1970s 1980s 1990s 2000s

Figure 5: Recent uptick in V + that (13 verbs, raw frequencies).

document) are everyday, colloquial words, mostly from a different register to those attested in my student essays. There isn’t enough information to mount a convincing constructional analysis that would explain specific rises in frequency or accretions to the class of verbs in V + that, when other verbs show no increase of usage or even (consider, reflect) a decline as we approach the present day. V + that is a long-established and semantically compositional construction, a member of the more abstract transitive schema. In discourse terms, V + that and shell + that can convey much the same,16 but V + that is more economical. It does not seem surprising that its distribution should widen. I was to some extent able to test my own impressions for signs of the Recency Illusion, having casually assumed some examples to be new, whether classified as error or innovation. Some clearly were not new at all, for example reflect + that. Others may show an Anglo-American difference: an obvious candidate here is like + that (though that pattern had not appeared in my students’ writing). Some patterns of V + that do seem to be recent, however, even in American English. For example, criticise + that, one of the usages that had struck me as new, occurs only once in COHA as far as I am aware (not, as it happens, in a pattern that my Python

16 They are semantically similar if the shell noun is appropriately general in meaning, as in express the opinion that ≈ express that, underscore the fact that ≈ underscore that, etc.

That-clauses as complements of verbs or nouns 

 79

script picks up), and the date is indeed recent. Likewise highlight + that occurs only twice, again in the same decade. (35)

The shop teacher criticized that the two bookends were not perfectly symmetrical and gave him a B for the project (2002, COHA)

(36)

The renowned Millennium Ecosystem Assessment […]highlights that coasts are the most highly degraded ecosystems on our planet. (2005, COHA)

4 Whose grammar? At some point in the investigation, explicit grammaticality judgments come into play, for example in excluding outright errors. However, such judgments are not always reliable. Although grammaticality judgments, even of sentences in isolation, can often achieve an uncontroversial consensus, outcomes may be influenced amongst other things by context, priming, dialect, idiolect, register, prescriptive ideas, and the Observer’s Paradox. Much of this applies to linguists as to anyone else. By definition, grammaticality must also figure in language change, since advanced and conservative speakers will differ in their (usually implicit) judgments. How well-defined, then, is grammaticality? This is much discussed. To give three examples among many, Ross (1973: 190–194 [2004: 389–391]) was an early commentator on the variation of acceptability of ‘squishy categories’ among speakers, and the extent to which idiolectal variation can be generalized into what could be called dialects. More recently, Mindt (2002) argued that a corpus investigation of a grammatical rule will typically find up to 5% of exceptions due to language change, linguistic creativity, intended ungrammaticality, etc. (cited by Mukherjee 2005: 100–1); note that our problem is in part lexical, which Mindt excludes from his observation. As for Sampson & Babarczy (2014), they dispute the validity of the whole grammatical ~ ungrammatical distinction. Within generative grammar there is a tradition of regarding graded judgments, whether from an individual speaker or the outcome of averaging the reports of many speakers, as a matter ‘merely’ of performance or E-Language and not part of the underlying grammar, which is assumed to be discrete and clearcut (Schütze 1996, 2004). If gradience is acknowledged at all, then, it is what happens when subjective, extra-grammatical factors cloud the ability of speakers to report their own grammatical knowledge reliably. The usage-based tradition, on the other hand, suggests that grammatical knowledge is inherently probabilistic (Gahl & Garnsey 2006, Gahl & Yu 2006,

80 

 David Denison

Bresnan 2007), which in some cases would disfavor or even preclude clear-cut judgments of grammaticality (see also introduction, this volume). I am sympathetic to this position, and indeed I regard speaker uncertainty as a useful diagnostic of usages that are marginal, problematic or undergoing change. I return to the synchronic register contrast between conversation and academic prose mentioned near the start of Section 1. It is explained as follows by Biber et al. (1999: 650). […] the expression of stance is backgrounded in noun complement clauses. With verb complement clauses, the subject of the verb is often a human agent or experiencer, so that the stance reported by the verb can be attributed directly to that person. In contrast, the stance conveyed by a controlling head noun is not normally attributed to anyone, so that readers must infer that the noun reports the stance of the writer. The opposite distributions of verb complement clauses (preferred in conversation) and noun complement clauses (preferred in academic writing) can thus be attributed to two factors. First, conversation has an overall preference for verbal rather than nominal structures, while academic prose shows the opposite pattern, preferring to integrate information in noun phrases. Second, the differing primary purposes of conversation and academic prose are at least as important here: conversational participants are interested in each others’ [sic] personal feelings and attitudes […] In contrast, academic writers are generally much more interested in the information being conveyed than personal attitudes […]

In the case of what I have called ‘unexpected V + that’, my limited experience of asking students for a grammaticality judgment on what they had submitted is that they did not have strong intuitions either way. (Admittedly, given the situation, the Observer’s Paradox may have contributed!). As unpracticed writers, they may have been extending a verbal constructional pattern more typical of conversation to an academic context, but inserting lexical items into the construction that would be rare or non-existent in their everyday conversation. Do they really have control of the usage? To what extent is sporadic occurrence of ‘new’ V + that patterns a part of their grammar? It is almost as if they are not native speakers of the register of English they are learning to use. We can compare this sort of innovation with what would be regarded as errors in advanced L2 usage; there are affinities also with the development of new Englishes. See Hackert (2009) on the history and politics of the term ‘native speaker’, as contrasted with ‘proficient speaker’ (Mufwene 1998). This line of thinking is inconvenient for the idea of a speaker having one grammar, and of that grammar being well defined. It suggests too that research on L2 learning is relevant to this piece of diachronic research. A more typical diachronic account would simply observe that some errors turn out with hindsight to have been innovations. Compare, say, the use of criteria as a singular instead of the ‘correct’ criterion. Evidently this represents a minor morpholexical change going on in the language. Similarly, the subject of

That-clauses as complements of verbs or nouns 

 81

this paper could also be seen as an unremarkable, language-internal extension of subcategorization. I contend that the latter viewpoint is not mutually exclusive with the previous discussion of registers and language learning.17 Our understanding of the V + that development can in principle be enhanced by examining the social, technological and historical context and by considering register and medium. Perhaps, somewhat naively, I had not been sufficiently aware of the probable chronological difference between American and British developments before collecting data from COHA.18 It must be admitted that the COHA data are rather different from the British student writing I had first observed and offer little support to the hypothesis of ‘change by thesaurus’.

5 What else? The wider historical context to the spread of V + that is that-clause complementation in general, which, as far as I know, is discussed mainly in relation to variation in two areas. The first is replacement by non-finite complements, specifically infinitives and -ing clauses (Rohdenburg 1995, Los 2005). Here the that-clause is the recessive variant, whereas we have been looking at a case where the thatclause qua verbal complement is the innovation. The difference, I suggest, is that non-finite clauses often involve a control or raising relationship between positions in higher and lower clause, whereas the verbs discussed in this paper are not open to subject raising, since they have agentive subjects, while the discourse function discourages a control relationship between higher and lower clause constituents, as the higher clause subject is not generally presenting information about, or persuasion directed at, themself. The second area of variationist research is the choice between that and zero (Elsness 1984, Rissanen 1991, Thompson & Mulac 1991, Finegan & Biber 1995, López-Couso 1996). I have encountered very few examples of zero marking with the verbs I have been studying – what we might call V + ∅ as a variant of V + that  – unsurprisingly, since my research method specifically targeted that. 17 Register and domain even have a bearing on a ‘simple’ change like criterion/criteria, as that lexical item has arguably been moving from educated usage into such specialized discourses as computing and employment practice, and from there back to more general use. 18 When I presented on this topic at LMEC6 in Uppsala, Terry Walker (British) reported having been struck by the frequency of ‘unexpected’ V + that while reading American Young Adult fiction, while Anne Curzan (American) observed that many of my ‘unexpected’ examples of V + that seemed quite normal to her.

82 

 David Denison

I suspect that the presence of V + ∅ could serve as a handy indirect indication of the prior entrenchment of V + that with the verb in question. This needs testing with a parsed corpus in order to seek out zero-marked clauses. For completeness we should mention the general distribution of shell + that – called ‘that-noun complement clause’ in Biber et al. (1999: 645–51). According to a more recent diachronic study from 1800 to the present, shell + that seems to be relatively infrequent by comparison with some other types of post-nominal modifiers and complements and yet to ‘have shown a strong increase in use, especially in academic writing’ (Biber & Gray 2012: 325). Such a finding need not contradict our results, since the replacement of shell + that by V + that is mostly recent, modest in numbers, and concentrated in particular verbs and perhaps in registers other than academic prose. This paper has raised more questions than it can answer, and good answers would need more time and resources than the questions perhaps deserve. Even within the corpus linguistic approach there is a need to add sensitivity to genre and register, to develop more sophisticated search algorithms, to use larger historical datasets than COHA, and to systematically examine the Corpus of Current American English (COCA) and other contemporary corpora. Future work could perhaps take advantage of the better search facilities of BNCweb and compare BNC with BNC2014 (released in September 2017) for a recent snapshot of change, and in British English too. On genre it would be helpful to compare a corpus of student writing such as the Michigan Corpus of Upper-Level Student Papers (MICUSP) with standard corpora of (mainly) copy-edited material, in order to see whether the student work shows greater variability or innovation in V + that. I am grateful to Elizabeth Traugott for drawing my attention to the study by Biber & Gray (2012): their ideas on economy and popularization may well be relevant here. The influence of technology on language is another area of research that could be brought to bear. Finally, and this is really a major desideratum, we need a psycholinguistic investigation of speakers’ knowledge in order to put the discussion sketched in Section 4 on a sounder footing.

Acknowledgements: Early versions were presented at ICLCE3 (London, 2009) and ISLE2 (Boston MA, 2011) and summarized in lectures at Aoyama Gakuin University (Tokyo, 2016) and the University of Hiroshima (2016). The statistics here were first presented at LMEC6 (Uppsala, 2017). I thank all those audiences, two anonymous referees and the series editor, Elizabeth Traugott, for a number of insightful comments. I am deeply grateful to Jiří Zámečník, a member of research training group “Frequency effects in language” at Freiburg, who generously provided the basis

That-clauses as complements of verbs or nouns 

 83

of two Python scripts for the offline COHA files. He helped to debug and improve these scripts, optimizing them to run 40 times faster than at first and to handle errors encountered in the files. He also pointed out the unreliability of a conclusion based on low frequencies.

References Aarts, Bas, David Denison, Evelien Keizer & Gergana Popova (eds.). 2004. Fuzzy grammar: A reader. Oxford: Oxford University Press. Biber, Douglas & Bethany Gray. 2012. The competing demands of popularization vs. economy: Written language in the age of mass literacy. In Terttu Nevalainen & Elizabeth Closs Traugott (eds.), The Oxford handbook of the history of English (Oxford Handbooks in Linguistics), 314–28. New York: Oxford University Press. Biber, Douglas, Stig Johansson, Geoffrey Leech, Susan Conrad & Edward Finegan. 1999. Longman grammar of spoken and written English. Harlow: Pearson. Bresnan, Joan. 2007. Is syntactic knowledge probabilistic? Experiments with the English dative alternation. In Sam Featherston & Wolfgang Sternefeld (eds.), Roots: Linguistics in search of its evidential base (Studies in Generative Grammar 96), 75–96. Berlin and New York: Mouton de Gruyter. Davies, Mark. 2010–. COHA: The Corpus of Historical American English: 400 million words, 1810–2009. Available online at http://corpus.byu.edu/coha/. Brigham Young University. De Smet, Hendrik. 2012. The course of actualization. Language 88.3. 601–33. Denison, David. 2009. A new class of verbs taking that-clause complements. Paper presented at ICLCE3, Symposium on Current Change in the English Verb Phrase, London. Denison, David. 2011. ISLE highlights? Paper presented at ISLE2, Boston MA. Denison, David. 2017. That-clauses as complements of verbs or nouns. Paper presented at The Sixth International Conference on Late Modern English (LMEC 6): Internal and External Factors in Linguistic Stability and Language Change, Uppsala. Elsness, J. 1984. That or zero? A look at the choice of object clause connective in a corpus of American English. English Studies 65. 519–33. Finegan, Edward & Douglas Biber. 1995. That and zero complementisers in Late Modern English: Exploring ARCHER from 1650–1990. In Bas Aarts & Charles Meyer (eds.), The verb in contemporary English, 241–57. Cambridge, etc: Cambridge University Press. Flowerdew, John & Richard W. Forest. 2015. Signalling nouns in English: A corpus-based discourse approach. Cambridge: Cambridge University Press. Gahl, Susanne & Susan Garnsey. 2006. Knowledge of grammar includes knowledge of syntactic probabilities. Language 82. 405–10. Gahl, Susanne & Alan C. L. Yu. 2006. Introduction to the special issue on exemplar-based models in linguistics. The Linguistic Review 23. 213–6. Hackert, Stephanie. 2009. A discourse-historical approach to the English native speaker. In Thomas Hoffmann & Lucia Siebers (eds.), World Englishes – Problems, properties and prospects: Selected papers from the 13th IAWE conference (Varieties of English Around the World G40), 385–406. Amsterdam and New York: John Benjamins.

84 

 David Denison

Hornby, Albert Sydney. 2015. Oxford advanced learner’s dictionary of current English, 9th edn. Oxford: Oxford University Press. http://www.oxfordlearnersdictionaries.com/definition/english/. Huddleston, Rodney & Geoffrey K. Pullum. 2002. The Cambridge grammar of the English language. Cambridge: Cambridge University Press. Hunston, Susan & Gill Francis. 2000. Pattern grammar: A corpus-driven approach to the lexical grammar of English (Studies in Corpus Linguistics 4). Amsterdam: John Benjamins. López-Couso, María José. 1996. A look at that/zero variation in Restoration English. In Derek Britton (ed.), English historical linguistics 1994: Papers from the 8th International Conference on English Historical Linguistics (8.ICEHL, Edinburgh, 19–23 September 1994) (Current Issues in Linguistic Theory 135), 271–86. Amsterdam and Philadelphia: John Benjamins. Los, Bettelou. 2005. The rise of the to-infinitive. Oxford: Oxford University Press. Mindt, Dieter. 2002. What is a grammatical rule? In Leiv Egil Breivik & Angela Hasselgren (eds.), From the COLT’s mouth … and others: Language corpora studies – in honour of Anna-Brita Stenström (Language and Computers – Studies in Practical Linguistics 40), 197–212. Amsterdam and New York: Rodopi. Mufwene, Salikoko. 1998. Native speaker, proficient speaker and norms. In Rajendra Singh (ed.), The native speaker: Multilingual perspectives, 111–23. New Delhi: Sage. Mukherjee, Joybrato. 2005. English ditransitive verbs: Aspects of theory, description and a usage-based model. Amsterdam: Rodopi. Newmeyer, Frederick J. 2003. Theoretical implications of grammatical category-grammatical relation mismatches. In Elaine J. Francis & Laura A. Michaelis (eds.), Mismatch: Form-function incongruity and the architecture of grammar (CSLI Lecture Notes 163), 149–78. Stanford: University of Chicago Press for CSLI Publications. Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech & Jan Svartvik. 1985. A comprehensive grammar of the English language. London and New York: Longman. Rissanen, Matti. 1991. On the history of that/zero as object-clause links in English. In Karin Aijmer & Bengt Altenberg (eds.), English corpus linguistics: Studies in honour of Jan Svartvik, 272–89. London and New York: Longman. Rohdenburg, Günter. 1995. On the replacement of finite complement clauses by infinitives in English. English Studies 76. 367–88. Ross, John Robert. 1973. Nouniness. In Osamu Fujimura (ed.), Three dimensions of linguistic theory, 137–257. Tokyo: TEC for Tokyo Institute for Advanced Studies of Language. Repr. In Aarts et al. (2004), 351–422. Sampson, Geoffrey & Anna Babarczy. 2014. Grammar without grammaticality: Growth and limits of grammatical precision (Topics in English Linguistics (TiEL) 254). Berlin and Boston: De Gruyter Mouton. Schmid, Hans-Jörg. 2000. English abstract nouns as conceptual shells: From corpus to cognition (Topics in English linguistics 34). Berlin and New York: Mouton de Gruyter. Schütze, Carson T. 1996. The empirical base of linguistics: Grammaticality judgments and linguistic methodology. Chicago and London: Chicago University Press. Schütze, Carson T. 2004. The nature of graded judgments. In Aarts et al. (2004), 431–46. Repr. of Section 3.3 of Schütze (1996). Thompson, Sandra A. & Anthony Mulac. 1991. The discourse conditions for the use of the complementizer that in conversational English. Journal of Pragmatics 15. 237–91. Zwicky, Arnold. 2005. Just between Dr. Language and I. Language Log. http://itre.cis.upenn. edu/~myl/languagelog/archives/002386.html (accessed 4 May 2006).

Juhani Rudanko

Semantic roles as a factor affecting complement choice: a case study with data from COHA Abstract: This article examines to infinitive and to -ing complements of the adjective unaccustomed using the entire COHA corpus. A first objective is to obtain information on the incidence of the two types of complements in recent English, and on whether there has been a change in their distribution over the last two centuries. A related question is whether any detectable change has been in accordance with the Great Complement Shift. The main focus of the study is then on whether semantic roles are a factor impacting the selection of the two types of complement. The hypothesis involves what in very recent work has been termed the Choice Principle, focusing on this principle in the case of the adjective unaccustomed, a matrix predicate that has not previously been considered from this perspective.

1 Introduction In the words of Huddleston and Pullum, the “most important property of complements in clause structure is that they require the presence of an appropriate verb that licenses them” (2002: 219). It seems possible to view Huddleston and Pullum’s concept of licensing as amounting to selection, and to say that a particular head – verb, adjective, or noun – selects a certain type of complement. It is then also possible to say that a complement is a phrasal constituent selected by a head. A basic task in the study of complementation is to provide an account of the factors that bear on complement selection. The purpose of this article is to contribute to this area of research. Complements may be non-sentential, as in eager for success, where the adjective eager selects a for NP complement, or they may be sentential. This study deals with two types of sentential complements, chosen because they have been felt to be close to each other by grammarians in earlier work. To introduce the constructions in question, consider the sentences in (1a–b), from the Corpus of Historical American English (COHA). Juhani Rudanko, Department of English, University of Tampere https://doi.org/10.1515/9783110583571-005

86 

 Juhani Rudanko

(1) a. I was eager to hear his news, … (1996, FIC) b. I am addicted to making strange friendships. (1896, FIC) In (1a) the complement of the adjective eager is a to infinitive clause. As for (1b), the complement of the adjective addicted is a gerundial complement consisting of the word to and a following -ing clause. The term to -ing complement is used here for this type of complement. An assumption made here concerns the postulation of understood subjects in sentential complements, as in (1a–b). Major traditional grammarians, including Otto Jespersen ([1940] 1961: 140) made this assumption, and it is also made in more recent work (see for instance Chomsky 1986: 114–131 and Davies and Dubinsky 2004: Chapter 1). The postulation of an understood subject makes it possible to represent the argument structure of the lower predicate in a straightforward way, with the understood subject representing the subject argument of the lower predicate. Another important similarity between the sentences of (1a–b) is that in both the higher subject receives a theta role1 from the higher predicate, and as a consequence we are dealing with a control construction in each case. In current work the type of understood subject is then represented by the symbol PRO, which stands for an abstract pronominal NP that is not pronounced (Chomsky 1981: 6; Davies and Dubinsky 2004: 84). Both (1a) and (1b) involve the word to, but it is assumed here that in current English the two types of to are different. The to linked to infinitives may be termed infinitival to, and the to that precedes an -ing clause is the preposition to. Infinitival to is placed under the Aux (or Infl) node for instance in Chomsky (1981: 18–19), and this analysis is adopted here. Prepositional to is of course placed under a PP node, and when the traditional notion of a nominal clause is used, it is possible to say that the PP node is rewritten as P – realized as prepositional to – and as a NP, with the NP dominating a S node. Taking the above remarks into account, it is possible to represent the sentences in (1a–b) in their relevant parts with the bracketing in (1a’-b’). (1) a.’ [[I]NP was [[eager]Adj [[PRO]NP [to]Aux [hear his news]VP]S2]AdjP]S1 b.’ [[I]NP am [[addicted]Adj [[to]Prep [[[PRO]NP making strange friendships]S2]NP]PP]AdjP]S1

1 For the purposes of the present investigation, theta roles can be viewed as “semantic roles played by arguments in relation to their predicates” (Radford 1997: 326).

Semantic roles as a factor affecting complement choice 

 87

The key question at this point is what syntactic evidence there is to separate the two types of to. VP Deletion has been considered as the “best piece of evidence” for placing to under the Aux label by Warner (1993: 64; see also Radford 1997: 53–4) and it is worth mentioning here. (Warner 1993: 5 actually argues that a better name for the construction is “post–auxiliary ellipsis,” but the more traditional label “VP Deletion” is retained here for the sake of convenience.) Consider the contrast between the sentences in (2a–b), constructed on the basis of the sentences in (1a–b). (2) a. I was eager to hear his news but my brother was not eager to. b. *I am addicted to making strange friendships, but my brother is not addicted to. The ellipsis rule can apply after an auxiliary, which explains why (2a) is well formed. In (2b) the word to is a preposition, and the corresponding ellipsis produces an ill-formed result, as expected. From a more general perspective, it is worth adding that in the case of the adjectives eager and addicted the two types of complement are not interchangeable, because the sentences in (3a–b) are both ill-formed in current English. The lack of interchangeability emphasizes the distinctness of the two complementation patterns. (3) a. *I was eager to hearing his news. b. *I am addicted to make strange friendships. Despite the sharp grammatical difference between the two types of to, there are matrix predicates that have shown variation and change between the two types of complement in recent times, including within the last two centuries. For instance, the adjective accustomed has been investigated from this perspective in a number of earlier studies, including Vosberg (2003a: 312–3 and Rudanko 2011: 100–154), with data from a number of large corpora. A number of verbs have also been studied with respect to the same variation, including consent (Rudanko 2012; 2017: Chapter 2). The general trend observed in earlier studies on this issue is that the gerundial complement has been spreading at the expense of the to infinitive (cf. Fanego 1996).2 The change in question is one major aspect of what has come 2 See Fanego (1996) for the period from 1400 to 1760 and Kjellmer (1980) and Denison (1998) for more recent change in this area. A comprehensive discussion of the emergence and spread of non-prepositional gerunds selected by matrix verbs from Middle English onwards is provided by de Smet (2013).

88 

 Juhani Rudanko

to be called the Great Complement Shift (see Rohdenburg 2006; Vosberg 2006; 2009; Fanego 2016; see also Rohdenburg, this volume).3 In earlier work it has been shown that the adjective accustomed has undergone a change in its complementation that has favored prepositional gerunds of the to -ing type at the expense of to infinitives. In the nineteenth century and at the beginning of the twentieth century to infinitive complements were more frequent than to -ing complements with accustomed, but over the course of the twentieth century to -ing complements became much more frequent than to infinitives with this adjective (Rudanko 2006; 2011: 100–154; Leech et al. 2009: 185–186). A first objective of the present article is to investigate the trajectories of the two types of non-finite complements with the adjective unaccustomed descriptively and to find out if they exhibit tendencies similar to those of accustomed in recent English. COHA is a large and balanced corpus, and it lends itself as a source of data for such an investigation. It would of course be desirable to conduct a similar investigation of British English, but that must be left to a future occasion when a comparable British English corpus becomes available. A further central issue highlighted in recent corpus-based work concerns the explanatory factors that may be identified as bearing on variation between infinitival and gerundial sentential complements, and this is a central issue in the analysis of the adjective unaccustomed in Section 2. This issue has more theoretical importance than a descriptive survey of complementation patterns, because any explanatory factors may also be tested with other matrix predicates in later work.

2 Nonfinite complements of the adjective unaccustomed in COHA As a preliminary to investigating data from COHA, it is appropriate to review briefly the treatment of the adjective unaccustomed in the Oxford English Dictionary (OED Online, December 2016), given the status of this dictionary as a standard work of reference. The relevant sense of the adjective is given as “Not accustomed or habituated” (sense 2.a in the OED), and under this sense the OED 3 The Great Complement Shift, as presented in Rohdenburg’s (2006) seminal study, covers a number of grammatical changes in English, including changes affecting to infinitives in relation to bare infinitives (Rohdenburg 2006: 144; see also Vosberg 2006 and Fanego 2016), but the emergence and spread of gerundial complements, both prepositional and non-prepositional, is a prominent feature of the Shift (Rohdenburg 2006; 143; Vosberg 2006: 17–19; Fanego 2016: 84–85).

Semantic roles as a factor affecting complement choice 

 89

mentions constructions with to, with illustrations of prepositional to and infinitival to side by side, without separating the two types of to. There are six examples, four featuring to NP complements and two featuring sentential complements. An example of a to NP complement is given in (4a), and the two sentential complements are given in (4b–c). (4) a. The abhorrence of bloodshed is common to all who are unaccustomed to it. (1846, A. Marsh) b. Your Heart, unaccustom’d to feel any very tender Impressions, felt some Concern for those you have inspir’d me with. (1728, E. Haywood) c. Familiar with crime, he was unaccustomed to be charged with it. (1891, F. W. Farrar) Both of the sentential complements are thus of the to infinitival variety, and there is no indication in the OED entry for unaccustomed that the adjective may take to -ing complements, and there are no examples of such complements among the illustrations of the entry for the adjective. On the other hand, the admissibility of what are here called to -ing complements has been noted in some other dictionaries, including the 7th edition of the Oxford Advanced Learner’s Dictionary of Current English (OALD 2005), but the lack of information on to -ing complements in the OED adds interest to the study of the adjective. COHA permits tagged searches and searches for to infinitive and to -ing complements of the adjective unaccustomed, and the search strings “unaccustomed to [v?i*]” and “unaccustomed to [v?g*]” were used as the basic search strings here. Despite their specificity, these strings retrieved some tokens that are not directly relevant, including the examples in (5a–b). (5) a. The soldiers, unaccustomed to exact discipline, had, in sport, set fire to a tall pine … (1835, FIC) b. … the irritable and eccentric Virginian, unaccustomed to defeat, displayed the conscious weakness of his cause … (1868, MAG) In both of (4a–b), the adjective of course selects a nonsentential to NP complement. The basic search strings were supplemented with search strings permitting one, two or three words between unaccustomed and to and further supplementary search strings were used that permitted one word between to and the infinitive and between to and the -ing form. The latter strings did not yield relevant tokens, but the former retrieved five relevant tokens. Three examples are given in (6a–c).

90 

 Juhani Rudanko

(6) a. … so unaccustomed was she to thinking in the first person … (1913, FIC) b. At dark they reached it [the ranch], the horses badly spent, unaccustomed as they were to moving faster than a walk. (1922, FIC) c. ... so unaccustomed was she to seeing an expression approaching tenderness on his face. (1993, FIC) Table 1 gives the numbers of tokens of each type of complement for the individual decades of the corpus, also indicating the size of the subcorpus for each decade. Table 1: To infinitive and to -ing complements of the adjective unaccustomed in the decades of COHA, with normalized frequencies per million words in parentheses. Decade

Size

1810s 1820s 1830s 1840s 1850s 1860s 1870s 1880s 1890s 1900s 1910s 1920s 1930s 1940s 1950s 1960s 1970s 1980s 1990s 2000s

1.2 6.9 13.8 16.0 16.5 17.1 18.6 20.3 20.6 22.1 22.7 25.7 24.6 24.3 24.5 24.0 23.8 25.3 27.9 29.6

to infinitives 4 (0.6) 8 (0.6) 3 (0.2) 9 (0.5) 8 (0.5) 6 (0.3) 8 (0.4) 5 (0.2) 1 (0.0) 7 (0.3) 1 (0.0) 1 (0.0)

1 (0.0)

to -ing

1 (0.1) 6 (0.4) 4 (0.2) 2 (0.1) 3 (0.1) 4 (0.2) 4 (0.2) 5 (0.2) 7 (0.3) 2 (0.1) 7 (0.3) 1 (0.0) 7 (0.3) 8 (0.3) 6 (0.2) 13 (0.5) 10 (0.3)

Initial illustrations of the two types of sentential complements from the present data are given in (7a–b), both from the same decade. (7) a. Moreover the compositors say that they are unaccustomed to compose in an unknown tongue from such scribbled and illegible copy, … (1911, NF) b. She was unaccustomed to doing her own hair and even the few days without a maid had given her no facility. (1914, FIC) The data in Table 1 indicate that both types of sentential complements have been found with the adjective unaccustomed in the last two centuries. They also

Semantic roles as a factor affecting complement choice 

 91

indicate that there has been a noticeable change. Throughout the nineteenth century to infinitives were more frequent than to -ing complements, and in the 1910s they were slightly more frequent than to -ing complements. However, to infinitives with unaccustomed are very rare in the later decades of the corpus, whereas to -ing complements are much more frequent. The change affecting the complement selection properties of unaccustomed is thus broadly similar to the trends that in earlier work (see Rudanko 2015: Chapter 2) have been found to affect the adjective accustomed, in that to infinitive complements have receded and there has been a change in favor of the gerundial complement. This is in accordance with a central feature of the Great Complement Shift. At the same time, the trends affecting unaccustomed and accustomed are not entirely identical. It was observed in Rudanko (2015: Chapter 2) on the basis of an analysis of the Fiction segment of COHA that to infinitives were still quite frequent (alongside of to -ing complements) with accustomed in the 1920s, 1930s and even in the 1940s, whereas the present study reveals that they were very rare with unaccustomed in those decades. Therefore, it appears that unaccustomed accommodated the gerundial pattern slightly earlier and perhaps slightly faster than accustomed, with the latter adjective being used much more frequently with non-finite complements overall. The data illustrates one example of a to infinitive complement from as late as the 1990s. It is given in (8). (8) … Dr. Kane is obviously a cultivated man, and by no means unaccustomed to watch the process of his own mind. (1997, NF) Sentence (8) is an isolated example from very recent English, but overall the data in Table 1 testify to a striking shift in favor of the to -ing pattern in relation to the to infinitive pattern. The change did not happen overnight, and the figures in Table 1 also show that it was during the period from the 1850s to the 1930s that there was at least some variation between the two types from one decade to the next. With this in mind, this period has been selected for closer examination here. The obvious analytical task is to inquire into the factors that may have a role in shedding light on variation between the two patterns of complementation, which, as noted in Section 1, are sharply different in terms of grammar. Two such factors are discussed here, the Extraction Principle and the Choice Principle.4 The former is 4 Structural discontinuities, of the type of (6a-c) in the text, have also been considered as a possible factor impacting complement choice, but as regards comparing the to infinitive complement with a prepositional gerund, it is not clear whether they are a factor (see Vosberg 2006: 175). As a consequence, the tokens in question – there are only three in the data from the period being

92 

 Juhani Rudanko

discussed only briefly, because it already has an established place in the literature, and more attention is paid to the Choice Principle, simply because it is novel and less established, and because it is of theoretical and methodological interest to examine the status of such a novel principle in the case of a matrix predicate that has not been investigated in this light before. Pioneering work on extractions in the context of variation between to infinitive and gerundial complements includes Vosberg (2003a) and (2003b) and Rohdenburg (2006), with Vosberg formulating the Extraction Principle. The Extraction Principle In the case of infinitival or gerundial complement options, the infinitive will tend to be favoured in environments where a complement of the subordinate clause is extracted (by topicalization, relativization, comparativization, or interrogation etc.) from its original position and crosses clause boundaries. (Vosberg 2003a: 308)5

Vosberg’s formulation refers to a “complement of the subordinate clause” being extracted, but in other work, including Vosberg (2006) and Rudanko (2006), account has also been taken of the extraction of adjuncts, and the principle is here used in the broader sense. The term “complement” is thus replaced by the term “constituent”. During the period from the 1850s to the 1930s there were two tokens with extraction. These are given in (9a–b). (9) a. Incidentally the shillings which the workingman himself was not unaccustomed to spend on drink? (1915, MAG) b. … extend even to the low latitudes where one is unaccustomed to see those golden draperies floating in the heavens. (1923, NF) Both examples involve Relativization and the complement in them is of the to infinitive type, as predicted by the Extraction Principle. The two tokens of extraction are set aside here in view of the well-established status of the Extraction

considered – have not been excluded when the Extraction Principle and the Choice Principle are examined. 5 In a wide-ranging study of extractions and of principles bearing on extraction contexts, Rohdenburg (2016) comes to the conclusion that the “marked infinitive [...] enjoys a privileged or target status in extraction contexts” (Rohdenburg 2016: 481). Of particular interest to the present study is his more specific finding that the marked infinitive, corresponding to the to infinitive as the term used in this study, “outranks all kinds of gerunds in such contexts” (Rohdenburg 2016: 481). This finding is in keeping with the more specific Extraction Principle used in the present article.

Semantic roles as a factor affecting complement choice 

 93

Principle. The low number of tokens with extractions means that the principle, while valuable, has a limited role in explaining variation in the present material. Turning to the Choice Principle, it is semantic in nature. The task of separating to infinitives and gerunds from the point of view of meaning has received considerable attention in the literature, and a large number of distinctions have been proposed. To give a flavor of such earlier work, Allerton’s (1988) approach is provided here. He argues that the “infinitive-gerund distinction, in its healthy state, can be summed up” with the notions in Table 2. Table 2: Allerton’s view of potential features bearing on the infinitive-gerund distinction. Infinitive

Gerund

infrequent activity intermittent activity interrupted activity uncompleted activity contingent / possible event particular time and place specific subject more verbal character

regular activity continuous activity continuing activity completed activity event presented factually neutral time and place non-specific subject more nominal character

(Source: Allerton 1988: 21)

The features in Allerton’s list do not always come to the surface; for instance, in the case of the specific versus nonspecific subject distinction, it seems clear that in both (7a) and (7b) the understood lower subjects of the predicates compose in an unknown tongue from such scribbled and illegible copy and doing her own hair are controlled by the higher subjects in a straightforward way. The distinctions in Allerton’s list, including the one just mentioned relating to control, are nevertheless worth bearing in mind. However, the Choice Principle approaches the infinitive versus gerund alternation in subordinate clauses from a different angle. The principle is formulated in Rudanko (2017: 20) as follows. The Choice Principle In the case of infinitival and gerundial complement options at a time of considerable variation between the two patterns, the infinitive tends to be associated with [+Choice] contexts and the gerund with [–Choice] contexts. A [+Choice] context is one where the lower predicate is agentive, assigning the Agent role to its subject, and a [–Choice] context is one where the lower predicate

94 

 Juhani Rudanko

is non-agentive and the lower subject is therefore not an Agent.6 (The terms [+/– Choice] are used here to refer to the interpretation of the lower subject and of the lower predicate, which assigns a semantic role to the lower subject.) To introduce the notion of agentivity independently of the predicate unaccustomed, consider the sentences in (10a–b), with the higher predicate accustomed, from this point of view. The sentences are from COHA. (See also Rudanko 2010: 12–13.) (10)

a. I am not accustomed to ordering my meals. (1916, FIC) b. Sir, I am not accustomed to having my word doubted. (1930, FIC)

The predicate ordering my meals is agentive in (10a), with its understood subject having the Agent role. There is probably no definition of agentivity or of the Agent role that would satisfy every investigator,7 but the approach here follows Rudanko (2011; 2015: 133–138) in drawing on Hundt (2004). She noted the three notions of volition, control8 and responsibility in her analysis of Agents and agentivity, and these notions are likewise emphasized here when making decisions on agentivity. In the approach adopted, an agentive predicate encodes a conceptualization of an event (or state) where the referent of the subject, whether explicit or 6 When conceptualizing or describing an event or a state of affairs, a language user (a speaker) has a range of options and means available, and the Choice Principle, as defined and used in this study, has to do with the interpretation of the means selected, that is, with the agentivity, or lack of it, of lower predicates and of their subjects. 7 Classic treatments of the Agent role and of agentivity include Gruber (1967), in whose view an “agentive verb is one whose subject refers to an animate object which is thought of as the willful source or agent of the activity described in the sentence” (1967: 943), and Fillmore (1968), who defines the Agentive as the “case of the typically animate perceived instigator of the action identified by the verb” (Fillmore 1968: 24). Some ten years later, Lakoff (1977: 244) considered “prototypical agent-patient sentences” and gave a long list of properties that may be found in them, including the volitionality of the agent’s action. For his part, Jackendoff (1990) identified “three semi-autonomous parts” in his discussion of the Agent role: “doer of action,” “volitional Actor,” and “extrinsic instigator” (1990: 129). The approach adopted in the present study is similarly based on identifying features of agentivity. It may be added that while there are verbs that typically involve an agentive interpretation, it is often insufficient to consider the verb of the sentence alone (see Marantz 1984: 23–25 and Chomsky 1986: 59–60), and the present approach also emphasizes that features of agentivity need to be assessed in the context of the use of each predicate. Sometimes it may also be helpful to consider the compatibility of an “intentional, agentive adverb” such as deliberately (see Givón 1993: 9) with a predicate in its context. To look back at (10a-b) in the text, such an adverb seems easily compatible with the predicate ordering my meals, but less so with having my word doubted, confirming the agentive status of the former. 8 As regards the notion of control relevant here, see also Berman (1970). It is of course different from the use of the word “control” in the grammatical term “subject control,” where it refers to the reference of PRO.

Semantic roles as a factor affecting complement choice 

 95

understood, is presented as being volitionally involved in the event in question,9 as exercising some degree of control over the event, and as being responsible for it. Thus in (10a) the predicate of the lower clause, ordering my meals, encodes an event such that the referent of the understood subject of the lower clause, coreferential with the higher subject, is volitionally involved in the event, exercises some control over it, and is responsible for it. By contrast, the predicate having one’s word doubted, as used in (10b), involves a conceptualization of an event where the referent of the understood subject is presented as neither volitionally involved in the event, nor in control of it, nor as responsible for it. In the case of (10b) the predicate suggests that the event happens to the referent of the understood subject, without the person being in control of it. In this case the subject is non-agentive. The syntactic pattern is of the type “have NP past participle,” and the same syntactic pattern may also have an agentive or causative interpretation, as in He had all the prisoners punished (from Palmer 1974: 199), but the interpretation of (10b) is not of the agentive type. The non-agentive interpretation in (10b) may be described as involving a “happenstance” have, and the example shows how the agentivity, or the lack of it, of a predicate should be assessed in the context of its use, rather than in an abstract or decontextualized manner. Overall, the notions of “undergoing and having something happen to one” (Thalberg 1967: 259), as illustrated in (10b), are helpful in understanding the nature of a non-agentive predicate. Imperatives, and the degree to which they are natural, may also shed light on the agentivity of a predicate or the lack of it. To understand the reason for this, consider this remark by John Taylor. Prototypically, an imperative instructs a person to do something, and is therefore only acceptable if a person has a choice between carrying out the instruction or not. (2003: 31)

From this perspective, the imperative Order your meals! is more likely and more natural than an imperative formed from the lower clause of (10b) Have your word doubted!, confirming the agentivity of the former. Some examples involving the adjective unaccustomed from the period under consideration, the 1850s to the 1930s, may be considered here to illustrate the distinction between [+Choice] and [–Choice] predicates further. Consider the examples in (11a–b) and (12a–b). (11)

a. You may not be aware of it, for you are unaccustomed to watch events which govern the future for good or evil; … (1871, FIC)

9 In his discussion of what he termed the Agent Proto-Role, Dowty (1991: 572) gave prominence to the notion of “volitional involvement in the event or state”. This captures the element of volition as an ingredient of agentivity.

96 

 Juhani Rudanko

b. Paul, who had worn off something of his diffidence, could not help feeling indignant at this speech; unaccustomed to be addressed in this way, the taunt jarred upon his feelings, but he only bit lip and preserved silence. (1865, FIC) (12)

a. There were marked inconsistencies in his face, but this was no disadvantage in a community unaccustomed to studying the external marks of character. (1866, FIC) b. I was so unaccustomed to having my work thrown back on my hands that I must have turned a little pale when I read the letter. (1895, FIC)

To assess the agentivity of a predicate, it needs to be considered in its context of use, as illustrated in the comments on example (10b) above. In (11a) and (12a) the relevant predicates are watch events which govern the future for good or evil and studying the external marks of character. These predicates, as used in their respective sentences, are [+Choice]. They encode volitional involvement, control and responsibility on the part of the understood subjects in their sentences, and an imperative of the type Watch events!, for instance, is entirely natural. By contrast, the predicates be addressed in this way and having one’s work thrown back on one’s hands are [–Choice]. The latter is of the happenstance have variety, discussed above, encoding what happens to the referent of the understood subject. As for example (11b), it illustrates another type of usage that is also typically [–Choice]. The lower clause is in the passive, and the understood subject corresponds to, or is derived from, the direct object of the corresponding active sentence, with that direct object having the Patient or Undergoer role. The Patient or Undergoer role is at the opposite end from the Agent role, and the construction illustrates a typical [–Choice] usage. The examples in (11b) and (12a) show immediately that the Choice Principle cannot be a categorical rule.10 However, to find out if the principle represents a noteworthy tendency in the present dataset, it is of interest to analyze the data from the 1850s to the 1930s from the point of view of the [+/–Choice] distinction. The results are given in Table 3.

10 The sentences in (11a-b) and (12a-b) also show that the [+/–Choice] distinction is independent of the [+/–Animate] feature of the lower subject. The adjective unaccustomed tends to select [+Animate], and even [+Human], subjects, with the lower subjects carrying the same features under coreference. However, to take (11a-b) as examples, the lower subjects are [+Animate] in both, but in (11a) the lower subject is [+Choice], and in (11b) it is [–Choice].

Semantic roles as a factor affecting complement choice 

 97

Table 3: To infinitive and to -ing complements of the adjective unaccustomed in [+Choice] and [–Choice] environments from the 1850s to the 1930s. Decade

1850s 1860s 1870s 1880s 1890s 1900s 1910s 1920s 1930s

to infinitives

to-ing

[+Choice]

[–Choice]

[+Choice]

[–Choice]

8 6 4 7 5 1 5

1 2 2 1

4 2 1 1 3 1 4 3 1

2 2 1 2 1 3 1 4 1

1

1

The results from Table 3 are summed up in Table 4. Table 4: To infinitive and to -ing complements with [+Choice] or [–Choice] readings during the period from the 1850s to the 1930s.

to infinitives to-ing

[+Choice]

[–Choice]

37 20

7 17

Applying the Chi-square test yields a Chi square of 7.32, and the results are statistically significant at the level of p whose >whom),10 the data in Figure 2, which contain all S1A files, support the following order: who >that >zero >which.11

10 This is also reinforced by results in Gut and Coronel (2012: 225) showing that to be the relativizer that scores highest in the overall account. 11 This difference may be accounted by the fact that Collins, Yao and Borlongan (2014) also included sentential relative clauses in their study, which justifies the higher frequent use of which than in my analysis.

A sociolinguistic study of relativizers in spoken Philippines English 

 297

So, the results from Figure 2 reveal important register differences in the selection of relativizers: while in written language invariable that dominates, and who scores third in frequency, in spontaneous conversation who is the most frequent choice and which lags behind. If we compare PhiE with the distribution in other Asian varieties (SuárezGómez 2014), the results show that PhiE shares with IndE the choice of who as the most favored relativizer, and with HKE and SgE the choice of the least favored relativizers, both zero and which. In Suárez-Gómez (2014) the prevalence of wh- words in IndE and of that in HKE and SgE was attributed to substrate influence. PhiE: who >that >zero >which IndE: who >which >that >zero HKE: that >who >zero >which SgE: that >who >zero >which The detailed analysis of PhiE in the rest of the section will shed light on its distribution of relativizers.

5.2 Variables determining the distribution of relativizers Among those factors said to have a significant influence on the choice of relativizer are the animacy of the antecedent, whether human or non-human, and the syntactic function of the relativizer. Who and whom combine with human antecedents, and which with non-human ones. Regarding that, zero and whose, these can make reference to both human and non-human antecedents. Although initially three variants were distinguished in this variable (“human”, “animate non-human” and “inanimate”), “animate non-human” and “inanimate” were conflated in the analysis into “non-human”, rendering the variable a binary one. The reason for this merging of variants is that only example (17) was categorized as animate non-human, and if we had to add a wh- relativizer, the most likely choice would by which, patterning with non-human or inanimate nouns. Regarding example (18), it was conflated with human antecedents, because of the presence of a human noun within the coordinating construction, which seems to determine the selection of who, although we are aware that the established correlation between who and human antecedents, and which and non-human ones, is not always held in World Englishes, as shown by Gut and Coronel (2012: 231) and Suárez-Gómez (2017) and indeed is also shown for PhiE in example (19) where which makes reference to the proper noun Vanessa.

298 

 Cristina Suárez-Gómez

(17) And if he couldn’t make anything it’d be one of the roses [Ø he grew in his tiny garden] (18) I think there’re only some production some production uh people or production houses [who has that parang I D nila] (19) And maybe my friend Vanessa [which is not yet sure if she’s coming] Table 1 illustrates the distribution of relativizers according to the semantic nature of the antecedent, whether human or non-human. Table 1: Distribution of relativizers according to the animacy of the antecedent.

Who Whom That Which Zero When, where, why, wherein, what TOTAL

Human

Non-human

TOTAL

265 (84.4%) 5 (1.6%) 22 (7%) 1 (0.3%) 21 (6.7%) 0 314

0 0 222 (36.3%) 85 (13.9%) 181 (29.6%) 123 (20.3%) 611

265 5 244 86 202 123 925

Table 1 shows a clear correlation between animacy of the antecedent and selection of relative markers in PhiE, especially with human antecedents, which select who(m) 86% of the time, showing a similar distribution to Present-day Standard English. That and zero with human antecedents are only used sporadically, unlike in Gut and Coronel (2012: 227) where a frequent use of zero with human antecedents is observed. Regarding non-human antecedents, the distribution is more balanced among the different possibilities and, although that is favored (36.3%), zero is also frequently used (29.6%), this also shown by Gut and Coronel (2012: 227). Unlike the case of human antecedents, where wh- words are favored, with non-human ones, which is surprisingly the least frequent choice (13.9%). Regarding when, where, why and wherein, these always combine with inanimate antecedents in adverbial function. Another factor that is repeatedly noted as having a significant influence on the selection of relativizers is the syntactic role within the relative clause. Grammars of contemporary English disfavor the use of zero, whom and whose in the subject slot, and observe a tendency to use who as subject and that as object (Quirk et al 1985: 1250–1252; Huddleston and Pullum 2002: 1048–1049). In the compilation of

A sociolinguistic study of relativizers in spoken Philippines English 

 299

the database, the following syntactic functions were found: subject (cf. examples [18, 19]), direct object, as in example (17), indirect object (20), complement of pied-piped preposition (21), complement of stranded preposition (22), complement of absent or missing preposition (23), subject complement (24) and adjunct (of place, time, manner and reason) (25). (20) Am I the first person only person here in La Salle [that you’ve told this]

(21) it is it has a historical dimension because uh we are supposed to be going through through a stage a stage in history [in which we are no longer we no longer belong to the modern age] (22) How about the most recent what would be the most recent music [that you have listened to] (23) Or was that your mother [that you were going down the escalator from Glorietta Four] (24) Well well you know that’s the kind of person [Ø I am] (25) but there’s this place in Manila [wherein uhm all the people in that bar are all from Bacolod] For the analysis here, however, this category was also simplified and non-subject functions were distinguished between “objects”, which comprises direct, indirect and prepositional objects, and “adverbials”. Therefore, the variable became ternary with the values “subject”, “object” and “adverbial”, although a micro-analysis of individual contexts will be reported where relevant. Since the variable “human” vs. “non-human” rendered significant results, it was also added to the results of subjects and objects, in order to obtain a more detailed distribution of relativizers. Table 2 contains relative words in subject context according to human and non-human antecedents. Table 2 confirms previous research here, which revealed that the syntactic function of the relative word and the humanness of the antecedent have an effect on the choice of relativizer, not only in the reference varieties (Quirk 1957; Tottie 1997; Tagliamonte 2002) but also in Asian varieties of English (SuárezGómez 2014). PhiE shares the preference for who when it functions as subject and refers back to human antecedents, with an almost categorical preference

300 

 Cristina Suárez-Gómez

Table 2: Distribution of subject relativizers according to the animacy of the antecedent. Human

Non-human

TOTAL

Who That Which

255 (96.2%) 8 (3%) 1 (0.4%)

0 63 (51.6%) 59 (48.4%)

255 71 60

Zero TOTAL

1 (0.4%) 265

0 122

1 387

(96.2%) here over other choices, such as that or zero, these only sporadically attested. PhiE also shows a similar distribution to that observed in the same analyzed sample of Indian English (ICE-IND, S1A) (Suárez-Gómez 2014), which also selects who in 96.2% of cases. Regarding non-human antecedents, that is seen to be the most frequent option (51.6%), confirming the results of Table 1. However, when functioning as subject, which becomes a very strong competitor (48.4%), adding a new option to that observed in IndE, HKE and SgE (Suárez-Gómez 2014): which is clearly favored in IndE and that in HKE and SgE; in PhiE both that and which are frequently used. Also of note for both human and non-human antecedents is the use of zero in this context, as in example (26), not expected to occupy the subject slot according to the grammars of contemporary English, but found in dialectal varieties of English, as well as World Englishes. (26) In fact when I when I was in my freshman year we had this batchmate [Ø turned out to be a wacko] she turned out to be insane If we analyze the distribution of relative markers in object function, which comprises direct object, indirect object and prepositional object (27) according to the reference of the antecedent, we obtain the figures in Table 3. (27) So Jimmy what do you think about this issue [that McIntyre talks about in After Virtue ] Table 3 shows that zero becomes the most frequent option (41.7%) with human antecedents, followed by who(m) (31%) and finally that (27.1%), which becomes the least preferred option, and if favored with human nouns would only be in object function. This shows the relevance of taking syntactic function into account, since the distribution of relativizers functioning as subjects and objects is very different. Another important aspect illustrated in Table 3

A sociolinguistic study of relativizers in spoken Philippines English 

 301

is that with human antecedents, subject is the dominant syntactic function, in 82% of cases (265 examples from a total of 323), in agreement with Silverstein’s animacy hierarchy (1976: 140) which establishes a correlation between human agents and subjects. Here we might also hypothesize the possible influence of Filipino, which as noted in Section 2 only allows nominative relativizers. Table 3: Distribution of object relativizers according to the animacy of the antecedent.

Who That Which Zero Whom What TOTAL

Human

Non-human

TOTAL

10 (20.8%) 13 (27.1%) 0 20 (41.7%) 5 (10.4%) 0 48

0 120 (52.6%) 19 (8.3%) 86 (37.7%) 0 3 (1.3)%) 228

10 133 19 106 5 3 276

Regarding non-human antecedents, these are more frequently relativized as objects (65.1%, 228 out of 350 cases). In this function, that is the most popular choice (52.6%), followed by zero (37.7%); which in this scenario has the lowest presence at just 8.3%. Also noticeable is the sporadic use of what (28) as a variant in this context, which although rarely used is recorded as an object in adnominal clauses in other World Englishes (Suárez-Gómez 2014). (28) They really they can really make everything [what they want] Finally within syntactic function a third category was distinguished, which includes relativizers functioning as adverbials with four different meanings: place, time, manner and reason. Such a distinction has been made due to the different distribution of relativizers, as shown in Table 4. Adnominal relative clauses introduced by a relativizer functioning as an adverbial are given scant attention in descriptive grammars of contemporary English, with the exception of Quirk et al. (1985: 17–20) and Biber et al. (1999: 624–630) and, to a lesser extent, Huddleston and Pullum (2002). However, they are relevant in any study of relativization because there is a wide range of variation in the selection of relative words. One of these choices entails the use of a relative adverb (where, wherein, when, why) or a relative word (which, that or zero), usually functioning as the complement of a preposition, as in examples (29) and (30).

302 

 Cristina Suárez-Gómez

Table 4: Distribution of relativizers in adverbial function (place, time, manner and reason).

That Zero Which Why When Where Wherein TOTAL

Place

Time

Manner

Reason

TOTAL

8 (11.8%) 2 (2.9%) 4 (5.9%) N/A N/A 37 (54.4%) 17 (25%) 68

26 (21.3%) 55 (45.1%) 2 (1.6%) N/A 39 (32%) N/A N/A 122

6 (14.6%) 34 (82.9%) 1 (2.4%) N/A N/A N/A N/A 41

0 1 (3.6%) 0 27 (96.4%) N/A N/A N/A 28

40 92 7 27 39 37 17 259

(29) But I still I I remember the time [when I left the country by myself] (30) No I don’t they don’t have uh a dorm [in which you can stay] but they have a lounge Quantitative analyses which examine these constructions include Tottie (1997) for British English, Guy and Bayley (1995) for American English, Tottie and Rey (1997) for African American Vernacular English, and Suárez-Gómez (2015) for Asian Englishes. For both BrE and AmE, Tottie (1997) and Guy and Bailey (1995) observe certain collocational patters in the most frequent constructions. For Early African American English, Tottie and Rey (1997) observe that relativizer zero is the most frequent choice in this context. For Asian varieties, Suárez-Gómez (2015) discovered important similarities between IndE, HKE and SgE and standardized varieties of English. The three Asian varieties show restricted and formulaic uses of adverbial relative constructions, especially in the case of reason (the reason why) and manner (the way Ø). More variation is observed for locative and temporal meanings, and here Suárez-Gómez (2015) highlights the prevalence of whwords in IndE which contrasts with the prevalence of that and zero in HKE, and also in SgE, especially in time constructions and less markedly so in locative ones. The findings in Table 4 confirm the previously mentioned collocational patterns for reason (the reason why) and manner (the way Ø) and show more variation for the expression of place and time. For place, where is the favored choice in 55.4% of cases, and wherein, a form recorded in neither the reference varieties nor in the Asian varieties analyzed by Suárez-Gómez (2015), becomes the second variant (25%). In the case of time, zero is the preferred marker (45.1%), followed by when (32%) and that (21.3%). Which scores very low as an adverbial, irrespective of meaning. In sum, although PhiE shows a similar distribution to other varieties in

A sociolinguistic study of relativizers in spoken Philippines English 

 303

the meanings of reason and manner, in terms of locative and temporal meanings it behaves differently from other Asian varieties, in the preference of wh- words for place, only shared with IndE, and zero and when for time. Of special interest are the relativizers functioning as complements of prepositions. Table 5 sets out the results here. Table 5: Distribution of relativizers as complements of preposition.

Who(m) Which That Zero Where TOTAL

Stranding

Pied-Piping

Absent

TOTAL

8 (14.9%) 2 (3.7%) 22 (40.7%) 21 (38.9%) 1 (1.9%) 54 (87.1%)

0 7 (100%) N/A N/A 0 7 (11.3%)

0 0 1 (100%) 0 0 1 (1.6%)

8 9 23 21 1 62

As in the Asian varieties analyzed by Suárez-Gómez (2014), namely IndE, HKE and SgE, the most frequent option in PhiE is preposition stranding (example [31]), which occurs in 87.1% of cases. Pied-piping also occurs in PhiE, but only on 7 occasions and in all of these with relativizer which, as in example (30), all functioning as adjuncts or adverbials. Curiously enough, the eight examples of who(m) functioning as a complement of a preposition are only attested with preposition stranding, and in all of them the prepositional phrase functions as a ‘prepositional object’, as in (32). (31) The pauses [that you were talking about] and of course my S’s I think are too pronounced (32) Well as far as I’m concerned they are the ones who the first time that I taught in this school they are the ones [whom I deal with] Together with preposition stranding and pied piping, a third possibility arises in the corpus, in which the relativizers function as the complement of a missing preposition, as in example (23), repeated for convenience as (33). (33) Or was that your mother [that you were going down the escalator from Glorietta Four] Here a stranded preposition (with) is deleted. Such structures seem to occur commonly in World Englishes, as pointed out by Newbrook (1998: 48–49) and Gut and Coronel (2012: 231). They are also recorded among the features listed by Kortmann

304 

 Cristina Suárez-Gómez

and Lunkenheimer (2011, feature 198) and according to Suárez-Gómez (2017) are justified as constructions derived from the learning process inherent to World Englishes. The results from PhiE are corroborated by the data from HKE and SgE in Suárez-Gómez (2014), where stranding is more frequent than pied-piping, but not in IndE, where preposition stranding and pied-piping show similar frequencies.

6 Summary and discussion This study has analyzed the distribution of relativizers in spontaneous conversations. The analysis is based on 925 examples of adnominal relative clauses introduced either by explicit relativizers or zero. Results show that in terms of absolute frequencies, relativizers in spontaneous conversation of PhiE rank as follows: who >that >zero >which. This is different from the written sample analyzed by Collins et al. (2014), in which the following arrangement was given: that >which >who >whose >whom. It is important to take into account that Collins et al. (2014) also included sentential relative clauses, which may justify the higher frequency of which than in my analysis of adnominal relative clauses. We can, then, tentatively conclude that PhiE shows important stylistic and register differences in the selection of relativizer, especially in the most frequent relativizer (who in spoken language vs. that in written language). When compared to the spoken register of other Asian varieties, spoken PhiE also shows idiosyncratic behavior, since it shares with IndE the choice of who as the dominant option, and with HKE and SgE the choice of which as the least frequent of the four. Regarding the contextual preferences of relativizers, who is the favored variant when referring back to human antecedents, and that if the antecedent is nonhuman. However, if we add the syntactic function of the relativizer to the animacy of the antecedent, we observe that with human antecedents, who remains the default – and almost the only – choice functioning as subject, whereas in object function zero is the favored option. In the case of non-human antecedents, that is the most frequent choice irrespective of syntactic function, yet as a subject which becomes a strong competitor, used almost as frequently as that. Finally, as adverbials, PhiE mirrors other Asian varieties in the collocational preferences of adverbials of reason, almost always introduced by why and manner, where zero is clearly favored, but not when indicating locative and temporal meanings. For time, zero is the most frequent relativizer, although when and that also occur frequently; regarding place, the adverbial relativizer where and wherein are prevalent, the latter a distinctive Philippine trait. Finally, if the relativizer functions as a complement of a preposition, the stranded construction dominates irrespective of the relativizer, similarly to HKE and SgE.

A sociolinguistic study of relativizers in spoken Philippines English 

 305

In addition to the quantitative analysis which reveals the distributional tendencies of relativizers in PhiE, the micro-analysis revealed the existence of structures which have not been accounted for in the parent variety, but have been reported in other World Englishes. We refer here to the use of which in combination with animate antecedents, the use of what introducing adnominal relative clauses, and constructions in which a stranded preposition is missing. These are accounted for as the result of cognitive processes of production in second-language settings, in this case manifested in the preference for simpler and more transparent structures. If we return to the hypotheses in Section 1, we can conclude that PhiE has accommodated to the general distribution of who observed in Present-day Standard English and in other Asian varieties of English, in that who is the default option as subject with human antecedents. However we cannot conclude that that supersedes which in the same syntactic function with non-human antecedents, as a consequence of the decrease of which at the expense of that (Leech et al. 2009: 227, 229); although it is used with a slightly greater frequency, the differences are not significant. Regarding the second and third hypotheses, the results of the analysis confirm the existence of innovative traces in relativization in PhiE, most of these also found in other L2 varieties and arising from the process of second language learning inherent to these varieties. With respect to the existence of structures or tendencies motivated by language contact, we have tentatively pointed out that the frequency of subject relative clauses might be attributed to the influence from Filipino, which only allows the relativization of nominatives; however, subject relative clauses are especially frequent in contexts of animacy, and this is also influenced by the universal tendency for animate subjects. Finally, the analysis of relative clauses in spoken language confirms that PhiE is developing its own endonormative tendencies and structures, as confirmed by the distribution of relativizers which differs from that reported for AmE, BrE or other Asian varieties, as well as by the existence of the wherein construction, only found in PhiE, and the availability of specific constructions commonly found in L2 varieties.

References Aldridge, Edith. 2004. Internally headed relative clauses in Austronesian languages. Language and Linguistics 5. 99–129. Alsagoff, Lubna & Ho Chee Lick. 1998. The relative clause in Colloquial Singapore English. World Englishes 17 (2). 127–138.

306 

 Cristina Suárez-Gómez

Bautista, M. Lourdes S. 2004. An overview of the Philippine component of the International Corpus of English (ICE-PHI). Asian Englishes 7 (2). 8–26. Bautista, M. Lourdes S. & Susan Butler (eds.). 2010. Anvil-Macquarie Philippine English Dictionary (Revised edn.). Pasig, the Philippines: Anvil Publishing, Inc. Biber, Douglas, Stig Johansson, Geoffrey Leech, Susan Conrad & Edward Finegan. 1999. Longman Grammar of Spoken and Written English. Harlow: Longman. Bolton, Kingsley. 2006. World Englishes today. In Braj B. Kachru, Yamuna Kachru & Cecil L. Nelson (eds.), The Handbook of World Englishes, 240–269. Oxford: Blackwell. Bolton, Kingsley & M. Lourdes S. Bautista. 2004. Philippine English: Tensions and transitions. World Englishes 23 (1). 1–5. Borlongan, Ariane M. 2009. A survey on language use, attitudes, and identity in relation to Philippine English among young generation Filipinos: An initial sample from a private university. Philippine ESL Journal 3. 74–107. Borlongan, Ariane M. 2016. Relocating Philippine English in Schneider’s dynamic model. Asian Englishes 18 (3). 232–241. Borlongan, Ariane M. & Shirley N. Dita. 2015. Taking a look at expanded predicates in Philippine English across time. Asian Englishes 17 (3). 240–247. Borlongan, Ariane M. & Joo-Hyuk Lim. 2013. Philippine English. In Bernd Kortmann and Kerstin Lunkenheimer (eds.), The Electronic World Atlas of Varieties of English. Leipzig, Germany: Max Planck Institute for Evolutionary Anthropology. Retrieved from the Electronic World Atlas of Varieties of English website http://www.ewave-atlas.org/ (accessed 20 September 2017). Burke, Isabelle. 2017. Wicked which: The linking relative in Australian English. Australian Journal of Linguistics 37 (3). 356–386. Collins, Peter. 2015. Recent diachronic change in the progressive in Philippine English. In Peter Collins (ed.), Grammatical change in English World-wide, 271–296. Amsterdam: John Benjamins. Collins, Peter, Ariane M. Borlongan, Joo-Hyuk Lim & Xinyue Yao. 2014. The subjunctive mood in Philippine English: A diachronic analysis. In Simone E. Pfenninger, Olga Timofeeva, Anne-Christine Gardner, Alop Honkapohja, Marianne Hundt & Daniel Schreier (eds.), Contact, Variation, and Change in the History of English, 259–280. Amsterdam: John Benjamins. Collins, Peter, Ariane M. Borlongan & Xinyue Yao. 2014. Modality in Philippine English: A diachronic study. Journal of English Linguistics 42. 68–88. Collins, Peter, Xinyue Yao & Ariane M. Borlongan. 2014. Relative clauses in Philippine English: A diachronic perspective. In Lieselotte Vandelanotte, Kristin Davidse, Caroline Gentens & Ditte Kimps (eds.), Recent Advances in Corpus Linguistics: Developing and Exploiting Corpora, 125–146. Amsterdam: Rodopi. Collins, Peter & Ariane M. Borlongan. 2017. Has Philippine English attained linguistic independence? The grammatical evidence. Australian ESL Journal 19. Coronel, Lilian. 2010. Relativization Strategies in Philippine English. Unpublished Master’s Thesis, University of Augsburg. Gil, David. 2013. Genitives, Adjectives and relative clauses. In Matthew S. Dryer & Martin Haspelmath (eds.), The World Atlas of Language Structures Online. Leipzig: Max Planck Institute for Evolutionary Anthropology. http://wals.info/chapter/60 (accessed 27 March 2017).  Gisborne, Nikolas. 2000. Relative clauses in Hong Kong English. World Englishes 19. 357–371. Gonzales, Wilkinson D. W. 2017. Philippines English. Asian Englishes 19 (1). 79–95.

A sociolinguistic study of relativizers in spoken Philippines English 

 307

Gonzalez, Andrew. 1982. English in the Philippine mass media. In John Pride (ed.), New Englishes, 211–216. Rowley: Newbury House. Gonzalez, Andrew. 2004. The social dimensions of Philippine English. World Englishes 23. 7–16. Greenbaum, Sidney. 1996. Introducing ICE. In Sidney Greenbaum (ed.), Comparing English Worldwide: The International Corpus of English, 3–12. Oxford: Clarendon Press. Gut, Ulrike & Lilian Coronel. 2012. Relatives worldwide. In Marianne Hundt & Ulrike Gut (eds.), Mapping Unity and Diversity worldwide: Corpus-based Studies of New Englishes, 215–242. Amsterdam: John Benjamins. Guy, Gregory R. & Robert Bayley. 1995. On the choice of relative pronouns in English. American Speech 70 (2). 148–172. Herrmann, Tanja. 2003. Relative Clauses in Dialects of English: A Typological Approach. Freiburg, Germany: University of Freiburg dissertation. Hinrichs, Lars, Benedikt Szmrecsanyi & Axel Bohmann. 2015. Which-hunting and the Standard English Relative Clause. Language 91 (4). 806–836. Huber, Magnus. 2012. Syntactic and variational complexity in british and Ghanaian English Relative clause formation in the written parts of the International Corpus of English. In Bernd Kortmann & Benedikt Szmrecsany (eds.), Linguistic Complexity, 218–242. Berlin and Boston: De Gruyter. Huddleston, Rodney & Geoffrey K. Pullum. 2002. The Cambridge Grammar of the English Language. Cambridge: Cambridge University Press. International Corpus of English. 1990–. Available through the ICE Project. http://ice-corpora. net/ice (accessed 27 March 2017). Keenan, Edward & Bernd Comrie. 1977. Noun Phrase accessibility and universal grammar. Linguistic Inquiry 8 (1). 63–99. Kortmann, Bernd & Kerstin Lunkenheimer (eds.). 2013. The Electronic World Atlas of Varieties of English [eWAVE]. Leipzig: Max Planck Institute for Evolutionary Anthropology. http://www. ewave-atlas.org/ (accessed April 4 2017). Kroeger, P. R. 1993. Phrase Structure and Grammatical Relations in Tagalog. Stanford: Center for the Study of Language and Information (CSLI) Publications. Leech, Geoffrey, Marianne Hundt, Christian Mair & Nicholas Smith, N. 2009. Change in Contemporary English. A Grammatical Study. Cambridge: Cambridge University Press. Levey, Stephen. 2006. Visiting London relatives. English World Wide 27. 45–70. Lewis, M. Paul, Gary Simons & Charles Fennig. 2016. Ethnologue: Languages of the World. Nineteenth Edition. Dallas: SIL International. Martin, Isabel P. 2014a. Beyond Nativization? Philippine English in Schneider’s Dynamic Model. In Sarah Buschfeld, Thomas Hoffmann, Magnus Huber & Alexander Kautzsch (eds.), The Evolution of Englishes: the Dynamic Model and Beyond, 70–85. Amsterdam: John Benjamins. McFarland, Curtis D. 2004. The Philippine language situation. World Englishes 23 (1). 59–75. Mesthrie, Rajend & Rakesh M. Bhatt. 2008. World Englishes: The Study of New Linguistic Varieties. Cambridge: Cambridge University Press. Miller, Jim. 2006. Spoken and written English. In Bas Aarts & April McMahon (eds.), The Handbook of English Linguistics, 603–624. Oxford: Blackwell. Mufwene, Salikoko S. 2014. Language ecology, language evolution, and the actuation question. In Tor Afarli & Brit Maelhum (eds.), Language Contact and Change: Grammatical Structure Encounters the Fluidity of Language, 13–35. Amsterfdam: Benjamins. Newbrook, Mark. 1998. Which way? That way? Variation and ongoing changes in the English relative clause. World Englishes 17 (1). 43–59.

308 

 Cristina Suárez-Gómez

Quirk, Randolph. 1957. Relative clauses in educated spoken English. English Studies 38. 97–109. Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech & Jan Svartvik. 1985. A Comprehensive Grammar of the English Language. London: Longman. Schachter, Paul & Otanes, Fé T. 1983. Tagalog Reference Grammar. Berkeley: University of California Press (reprinted edition). Schneider, Edgar W. 2007. Postcolonial English: Varieties around the World. Cambridge: Cambridge University Press. Schneider, Edgar W. 2011. The subjunctive in Philippine English: An updated assessment. In M. Lourdes S. Bautista (ed.) Studies of Philippine English: Exploring the Philippine Component of the International Corpus of English, 159–173. Manila: Anvil. Schneider, Edgar W. 2016. Hybrid Englishes: An exploratory survey. World Englishes 35. 339–354. Suárez-Gómez, Cristina. 2014. Relative clauses in Asian English. Journal of English Linguistics 42 (3). 245–268. Suárez-Gómez, Cristina. 2015. The places where English is spoken: adverbial relative clauses in World Englishes. World Englishes 34 (4). 620–635. Suárez-Gómez, Cristina. 2017. The role of transparency and language contact in the structural nativization of relative clauses in New Englishes. English World-Wide 38 (2). 212–238. Tagliamonte, Sali. 2002. Variation and change in the British relative marker system. In Patricia Poussa (ed.), Dialect contact on the North Sea Littoral, 147–165. Munich: Lincom Europa. Thomason, Sarah G. & Terrence Kaufmann. 1988. Language contact, creolization, and genetic linguistics. Berkeley and Los Angeles: University of California Press. Tottie, Gunnel. 1997. Overseas relatives: British-American differences in relative marker usage. In Jan Aarts, Inge de Mönnink & Herman Wekker (eds.), Studies in English Language and Teaching. In honour of Flor Aarts, 153–165. Amsterdam: Rodopi. Tottie, Gunnel & Dawn Harvie. 2000. It’s all relative: relativization strategies in Early African American English. In Shana Poplack (ed.), The English History of African American English, 198–230. Oxford: Blackwell. Tottie, Gunnel & Michael Rey. 1997. Relativization strategies in Earlier African American Vernacular English. Language Variation and Change 9 (2). 219–247 Yip, Virginia & Stephen Matthews. 1991. Relative complexity: beyond avoidance. Chinese University of Hong Kong Papers in Linguistics 3. 112–124.

Subject index academic prose 16, 62, 80, 82, 271, 273, 274, 275, 276, 277, 278, 281 agent role 93, 94, 96 American English 9, 11, 78, 105, 197, 216, 258, 285, 286, 287, 289, 302 analogy 4, 9, 10, 15, 47, 51, 55, 65, 66, 163, 170, 171, 172, 173, 238 animacy 287, 297, 298, 300, 301, 304, 305 antecedent 5, 16, 17, 18, 56, 263, 265, 268, 269, 270, 271, 272, 274, 277, 278, 279, 282, 285, 287, 289, 293, 297, 298, 299, 300, 301, 304, 305 apparent constituents 107, 125 attractor sets 3, 4, 13, 155, 163, 180 auxiliary-main verb scale 212, 213 backgrounding 1, 2, 8, 26, 28, 33, 34, 36, 37, 40, 80, 289–293 because of – as preposition 44 because X construction – because + Adj 45 – because + NP 9, 46 – causal because construction 9, 51, 52, 53, 56, 57 – pragmatics of 53 – semantics of 13, 15, 56, 161 British English 12, 15, 82, 88, 105, 129, 131, 137, 213, 215, 216, 219, 220, 221, 222, 225, 226, 227, 228, 229, 230, 231, 232, 233, 239, 243, 247, 285, 286, 302 British English-New English divide 226, 232 British National Corpus (BNC) 11, 25, 183 categorial incursion 4, 5, 13, 157, 178, 179, 180 catenative – catenative construction 2, 13, 204, 213, 217, 218, 221, 223, 224, 226, 228–229, 232, 233 – catenative get 14, 15, 185, 213, 214, 215, 225, 231, 233 causal meaning 253 causation 163, 180, 188, 200, 201 https://doi.org/10.1515/9783110583571-014

causative verbs 158, 160, 161, 163, 165, 166, 171, 172, 174, 180, 196, 197 Choice Principle 10, 85, 91, 92, 93, 94, 96, 97, 98, 99, 100 collocational profile 13, 161, 177, 189 complexity 1, 2, 5, 14, 47, 114, 120, 198, 204, 232, 244, 256, 272 Complexity Principle 11, 103, 115, 119, 124 compositionality 47, 188 compressed subordinate clauses 2, 7, 25, 28 computer-mediated communication 43, 53, 56 concessive conditionals 16, 263, 265, 266, 267, 268, 269, 271, 282 concessive connectives 265 concessive meaning 249, 250, 265 construal 6, 25, 33, 35, 37, 38, 40 constructionalisation 14, 186, 188 Construction Grammar (CxG) VIII, 3, 8, 9, 16, 43, 46, 47, 53, 204, 263, 264, 271, 278 contrastive meaning 240, 252 coreference 96 corpus linguistics 82 Corpus of Contemporary American English (COCA) 9, 13, 16, 43, 45, 46, 48, 125, 183, 186, 191, 194, 195, 196, 197, 204, 264, 271, 272, 274, 278 Corpus of Historical American English (COHA) 9, 10, 11, 13, 45, 46, 48, 61, 62, 68, 69, 70, 71, 73, 74, 75, 77, 78, 81, 82, 83, 85, 99, 105, 114, 118, 125, 186, 187, 189, 197, 204 Corpus of Late Modern English Texts (CLMET) 28, 32, 41 Czech National Corpus-Intercorp 199, 206 Diachronic Corpus of Present-Day Spoken English (DCPSE) 12, 70, 129, 131, 132, 133, 134, 137, 138, 139, 144, 147, 149, 150, 151 Double Object Construction 161 Early English Books Online 29, 35, 41 ellipsis 16, 17, 87, 271, 273, 275, 278, 279, 280

310 

 Subject index

endonormative 286, 290, 291, 292, 305 English as a Foreign Language (EFL) 15, 236, 243 English as a Native Language (ENL) 15, 16, 215, 230, 233, 236, 243, 258 English as a Second Language (ESL) 15, 214, 236, 242, 243, 245, 246, 248, 249, 250, 251, 252, 257, 258, 259 epistemic modality 16, 17, 269, 270, 271, 272, 273, 274, 278, 279, 282 expanded category/type/structure/ construction 11, 103, 105, 106, 107, 110, 114, 123, 124 external meaning 4, 157, 166, 180 Extraction Principle 10, 91, 92, 93 Freiburg-LOB Corpus (FLOB) 14, 28, 213, 214, 215, 225 from-gerund 11, 105, 109, 118, 122, 123, 124 from-phrases 122, 123 functional change 26, 27, 40, 41 functional expansion 32, 40 Fuzzy Tree Fragments (FTFs) 138, 139, 140, 141, 142, 143 German 14, 54, 185, 187, 190, 199, 200, 201, 202, 203, 204, 265, 274 gerund – gerundial common case subject 115 – gerund-participial clause 130 Global Web-based English corpus (GloWbE) 14, 15, 46, 213, 214, 215, 216, 225, 226, 227, 228, 229, 230, 231, 232, 233 Great Complement Shift 2, 10, 85, 88, 91

internal meaning 4, 159, 160, 166, 168, 169, 177, 180 International Corpus Network of Asian Learners of English (ICNALE) 15, 236, 237, 239, 240, 241, 244, 245, 247, 251 International Corpus of English Corpus Utility Program (ICECUP) 138, 139 International Corpus of English (ICE) 14, 15, 213, 214, 215, 225, 226, 227, 228, 229, 230, 231, 232, 233, 236, 237, 239, 240, 241, 244, 245, 247, 248, 250, 286 Lancaster-Oslo/Bergen Corpus (LOB) 14, 213, 216 language contact 286, 290, 292, 293, 305 London-Lund Corpus (LLC) 138, 144, 145, 146, 147, 149, 150, 153 mental representation verbs 155, 156, 159, 160, 161, 162 metaphorization 4, 5, 6, 13, 166, 167, 172, 174, 180 mode 219, 220, 231 Modern English 4, 7, 11, 12, 13, 25, 28, 29, 35, 39, 122, 124, 155, 156, 161, 185, 204 multi-source construction 13, 155, 157, 177, 178, 179, 180 nativization 241, 285, 290 negative implicative verbs 11, 103, 104, 105, 108, 109, 113, 118, 123, 124

hybrids/blends 104, 105, 107, 178

object case pronoun 11, 107, 109, 124 online communication. See computermediated communication; Oxford English Dictionary (OED) 10, 36, 38, 64, 88, 89, 126, 218

implicature 190 infinitive clauses 86, 133, 140, 143, 144, 151 -ing clauses 6, 12, 81, 153 -ing-form 192, 212, 213, 217, 218, 221, 223, 229, 230, 231, 233 innovative constructions 8, 214, 218, 223, 233 internalization 12, 13, 155, 156, 157, 161, 167, 179, 180

past participle 37, 38, 95, 231 Penn-Helsinki Parsed Corpus of Early Modern English (PPCEME2) 157, 158, 162, 183 Penn-Helsinki Parsed Corpus of Middle English (PPCME3) 157, 158, 162, 183 Penn Parsed Corpus of Modern British English (PPCMBE) 70, 157, 158, 162, 183 periphery 5, 266, 267, 270, 271, 272, 278, 279, 282

Subject index 

pied piping 303, 304 plagiarism 63 polysemization 4, 5, 13, 155, 157, 163, 170, 172 – polysemy 158, 236 possessive verbs 158, 166, 180 premodification 48 preposition stranding 303, 304 productivity 46, 155, 156, 162, 174, 178, 179, 180, 188 reflexive help -ing 11, 105, 107, 108, 111, 112, 113, 114, 115, 116, 117, 119, 120, 123, 124 reflexive pronoun 13, 106, 107, 108, 118, 125, 166, 167, 168, 169, 170, 172 relativizer 17, 69, 70, 285, 286, 287, 288, 289, 293, 294, 295, 296, 297, 298, 299, 300, 301, 302, 303, 304, 305 Romance influence 4, 157, 163, 174 Secondary Predicate Construction (SPC) – as-Secondary Predicate Construction (SPC) 156, 168 – for-Secondary Predicate Construction (SPC) 156 – (mental) zero-Secondary Predicate Construction 12, 13, 155, 156, 157, 158, 160, 161, 162, 163, 164, 165, 169, 170, 171, 172, 173, 174, 175, 177, 178, 179, 180 sentential complementation 66 shell noun 9, 10, 61, 62, 71, 72, 73, 74, 78 simplicity 14, 114, 232, 233 simultaneity 8, 25, 38, 39, 40, 236 since-clauses 16, 235, 236, 237, 239, 245, 246, 247, 248, 251, 252, 253, 254, 255, 256, 257, 258 situation-oriented present participles 2, 7, 25, 28, 30, 38, 39, 40 small clause 2, 4, 12, 155, 157 Spanish 8, 14, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 40, 187, 190, 199, 200, 201, 202, 203, 204, 289 spoken language 9, 14, 46, 54, 56, 58, 114, 211, 219, 220, 231, 241, 244, 245, 252, 253, 254, 258, 286, 289, 304, 305

 311

subcategorization 64, 65, 67, 81 subject control 94 subjectivity 5, 9, 44, 54, 56, 57, 279 substrate 6, 224, 233, 286, 292, 297 synonymy 53, 124 syntactic function 238, 285, 287, 297, 299, 300, 301, 304, 305 temporal meaning 238, 251, 252, 302, 303, 304 that-clause 6, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 73, 74, 81, 122 the adjacent possible 3, 5, 174 to-infinitive 1, 6, 122, 133, 187, 192, 194, 211, 212, 213, 217, 218, 221, 222, 223, 228, 229, 230, 233 transitivity 123 translation 8, 14, 25, 28, 29, 30, 31, 33, 35, 36, 37, 40, 164, 165, 169, 185, 186, 187, 198, 199, 200, 201, 202, 203, 204 Traugott’s Tendencies of semantic change 160 variation 10, 11, 43, 62, 79, 81, 87, 88, 91, 92, 93, 97, 98, 100, 117, 130, 223, 236, 263, 275, 286, 294, 301, 302 verbalization 2, 8, 12, 25, 26, 27, 28, 39, 152 verbal noun 12, 129, 130 verbs of communication 62, 166 verbs of negative causation 109, 113, 118, 122, 123, 124 verbs of perception 164 volition 6, 13, 94, 95, 96, 156, 188, 190, 196, 201 while-clauses 15, 16, 235, 237, 243, 248, 249, 250, 251, 252, 253, 254, 255, 257, 258

York-Toronto-Helsinki Parsed Corpus of Old English Prose (YCOE) 12, 157, 158, 162, 183