CURRENT TRENDS IN DIACHRONIC SEMANTICS AND PRAGMATICS
STUDIES IN PRAGMATICS Series Editors: Bruce Fraser, Kerstin Fischer, Maj-Britt Mosegaard Hansen
The Studies in Pragmatics series is dedicated to publishing innovative, authoritative monographs and edited collections from all micro-, macro-, and meta-pragmatic linguistic perspectives. Rooted in the interdisciplinary spirit of the Journal of Pragmatics, it welcomes not only book proposals from linguistics proper but also pragmatically oriented proposals from neighboring disciplines such as interactional sociology, language philosophy, communication science, social psychology, cognitive science, and information science. The goal of the series is to provide a widely read and respected international forum for high-quality theoretical, analytical, and applied pragmatic studies of all types. By publishing leading-edge work on natural language practice, it seeks to extend our growing knowledge of the forms, functions, and foundations of human interaction.
Other titles in this series: FISCHER
Approaches to Discourse Particles
AIJMER & SIMONVANDENBERGEN
Pragmatic Markers in Contrast
FETZER & FISCHER
Lexical Markers of Common Grounds
CAFFI
Mitigation
ROSSARI, RICCI & SPIRIDON
Grammaticalization and Pragmatics: Facts, Approaches, Theoretical Issues
FRASER & TURNER
Language in Life, and a Life in Language: Jacob Mey – A Festschrift
Proposals for the series are welcome. Please contact the Series Editor, Bruce Fraser:
[email protected]
CURRENT TRENDS IN DIACHRONIC SEMANTICS AND PRAGMATICS
EDITED BY MAJ-BRITT MOSEGAARD HANSEN The University of Manchester, UK
JACQUELINE VISCONTI University of Genova, Italy
United Kingdom – North America – Japan India – Malaysia – China
Emerald Group Publishing Limited Howard House, Wagon Lane, Bingley BD16 1WA, UK First edition 2009 Copyright r 2009 Emerald Group Publishing Limited Reprints and permission service Contact:
[email protected] No part of this book may be reproduced, stored in a retrieval system, transmitted in any form or by any means electronic, mechanical, photocopying, recording or otherwise without either the prior written permission of the publisher or a licence permitting restricted copying issued in the UK by The Copyright Licensing Agency and in the USA by The Copyright Clearance Center. No responsibility is accepted for the accuracy of information contained in the text, illustrations or advertisements. The opinions expressed in these chapters are not necessarily those of the Editor or the publisher. British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library ISBN: 978-1-84950-677-9 ISSN: 1750-368X (Series)
Awarded in recognition of Emerald’s production department’s adherence to quality systems and processes when preparing scholarly journals for print
Studies in Pragmatics (SiP) Series Editors Bruce Fraser Boston University, USA Kerstin Fischer University of Southern Denmark Maj-Britt Mosegaard Hansen University of Manchester, UK Consulting Editor Jacob L. Mey University of Southern Denmark Editorial Board Diane Blakemore, University of Salford, UK Shoshana Blum-Kulka, Hebrew University, Israel Laurel Brinton, University of British Columbia, Canada Claudia Caffi, University of Genoa, Italy Alessandro Duranti, UCLA, USA Anita Fetzer, University of Lueneburg, Germany Marjorie Goodwin, UCLA, USA Hartmut Haberland, University of Roskilde, Denmark William F. Hanks, University of California, USA Sachiko Ide, Tokyo Women’s University, Japan Kasia Jaszczolt, University of Cambridge, UK Elizabeth Keating, University of Texas, USA Sotaro Kita, University of Bristol, UK Ron Kuzar, University of Haifa, Israel Lorenzo Mondada, University of Lyon 2, France Henning Nølke, University of Aarhus, Denmark Etsuko Oishi, Fuji Women’s University, Japan Srikant Sarangi, Cardiff University, UK Marina Sbisa`, University of Trieste, Italy
TABLE
OF
CONTENTS
List of Contributors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
ix
Current Trends in Diachronic Semantics and Pragmatics . . . . . . . . . . . . . . . . . . . . . . . . Maj-Britt Mosegaard Hansen and Jacqueline Visconti
1
APO: Avoid Pragmatic Overload . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 Regine Eckardt Diachronic Pathways and Pragmatic Strategies: Different Types of Pragmatic Particles from a Diachronic Point of View. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 Ulrich Detges and Richard Waltereit Context Sensitive Changes: The Development of the Affirmative Markers godt ‘good’ and vel ‘well’ in Danish . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63 Eva Skafte Jensen Procatalepsis and the Etymology of Hedging and Boosting Particles . . . . . . . . . . . . . . . . 81 Kate Beeching Central/Peripheral Functions of allora and ‘Overall Pragmatic Configuration’: A Diachronic Perspective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107 Carla Bazzanella and Johanna Miecznikowski The Importance of Paradigms in Grammaticalisation: Spanish Digressive Markers por cierto and a propo´sito . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123 Maria Estelle´s The Multiple Origin of es que in Modern Spanish: Diachronic Evidence . . . . . . . . . . . . . 147 Magdalena Romera From Aspect/Mood Marker to Discourse Particle: Reconstructing Syntactic and Semantic Change. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165 Bethwyn Evans
viii
Table of Contents
The Grammaticalization Channels of Evidentials and Modal Particles in German: Integration in Textual Structures as a Common Feature . . . . . . . . . . . . . . . . . . . . . . . . . 189 Gabriele Diewald, Marijana Kresic and Elena Smirnova Evidentiality, Epistemicity, and their Diachronic Connections to Non-Factuality . . . . . . . . 211 Mario Squartini The Grammaticalization of Negative Reinforcers in Old and Middle French: A Discourse–Functional Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227 Maj-Britt Mosegaard Hansen A Roots Journey of a French Preposition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 253 Silvia Adler and Maria Asnes The Grammaticalization of Privative Adjectives: The Case of Mere. . . . . . . . . . . . . . . . . 273 Elke Gehweiler The Origin of Semantic Change in Discourse Tradition: A Case Study . . . . . . . . . . . . . . 291 Katerina Stathi
LIST
OF
CONTRIBUTORS
Silvia Adler, University of Haifa, Haifa, Israel Maria Asnes, Bar Ilan University, Ramat Gan, Israel Carla Bazzanella, Universita` degli Studi, Turin, Italy Kate Beeching, University of the West of England, Bristol, UK Ulrich Detges, Institute for Romance Philology, University of Munich, Germany Gabriele Diewald, Leibniz Universita¨t Hannover, Germany Regine Eckardt, University of Go¨ttingen, Go¨ttingen, Germany Maria Estelle´s, Universidad de Valencia, Spain Bethwyn Evans, Max Planck Institute for Evolutionary Anthropology, Leipzig, Germany Elke Gehweiler, Freie Universita¨t Berlin, Germany Maj-Britt Mosegaard Hansen, The University of Manchester, UK Eva Skafte Jensen, Roskilde University, Roskilde, Denmark Marijana Kresic, University of Zadar, Croatia Johanna Miecznikowski, Universita` della Svizzera italiana, Lugano, Switzerland Magdalena Romera, Universidad de las Islas Baleares, Palma de Mallorca, Baleares, Spain Elena Smirnova, Leibniz Universita¨t Hannover, Germany Mario Squartini, Universita` di Torino, Torino, Italy Katerina Stathi, Freie Universita¨t Berlin, Germany Jacqueline Visconti, University of Genova, Italy Richard Waltereit, Newcastle University, UK
Current Trends in Diachronic Semantics and Pragmatics Edited by Maj-Britt Mosegaard Hansen and Jacqueline Visconti r 2009 by Emerald Group Publishing Limited. All rights reserved.
1 CURRENT TRENDS AND PRAGMATICS
IN
DIACHRONIC SEMANTICS
Maj-Britt Mosegaard Hansen and Jacqueline Visconti
1. THE ROLE
OF
PRAGMATIC INFERENCING
IN
MEANING CHANGE
Research on semantic change has gained considerable momentum from the idea that pragmatic factors are a driving force in the process. The idea, first suggested by Grice (1989 [1975]: 39), that ‘‘it may not be impossible for what starts life, so to speak, as a conversational implicature to become conventionalized’’, was systematized in Traugott’s Invited Inferencing Theory of Semantic Change (IITSC) (Traugott, 1999; Traugott and Dasher, 2002). More precisely, semantic change is seen as ‘‘arising out of the pragmatic uses to which speakers or writers and addressees or readers put language, and most especially out of the preferred strategies that speakers/writers use in communicating with addressees’’ (Traugott and Dasher, 2002: xi). The model, which is based on Levinson’s theory of generalized conversational implicature (GCIs) (Levinson, 2000), in particular on the distinction between ‘‘coded’’, ‘‘utterance-type’’ and ‘‘utterance-token’’ levels of meaning (Levinson, 1995), is represented in Figure 1.1 (from Traugott and Dasher, 2002: 38). In this model, assuming that the meaning M1 of a lexeme L is linked to the conceptual structure Ca at stage I, innovation may be produced by speakers/writers employing L in a particular, ‘‘utterance-token’’, use. Should such a use spread to more contexts and become salient in the community, it acquires the status of a ‘‘generalized invited inference’’ and may eventually become semanticized as a new coded meaning M2 for L at stage II. The IITSC arose in the context of reflection on replicated cross-linguistic regularities in semantic change, which the theory explained by assuming that they are the outcome of similar cognitive and communicative processes across languages. This has a number of consequences,
2
Maj-Britt Mosegaard Hansen and Jacqueline Visconti
Figure 1.1. Model of the Invited Inferencing Theory of Semantic Change (IITSC; Traugott, 1999: 96) (M ¼ coded meaning; C ¼ conceptual structure) from Traugott and Dasher (2002: 38).
salient among which is the idea that meanings can be predicted to become increasingly pragmatic, procedural, and metatextual. More specifically, a significant number of observed meaning changes appear to instantiate the following clines, referred to as ‘‘semantic–pragmatic tendencies’’: (i) truth-conditionalWnon-truth-conditional; (ii) contentWcontent/proceduralWprocedural; (iii) scope-within-propositionWscope-over-propositionWscope-over-discourse; (iv) non-subjectiveWsubjectiveWintersubjective (Traugott and Dasher, 2002: 40).1
1
These clines represent a revision of three, by now very well known, tendencies posited in Traugott (1989: 34–35), viz. (I) Meanings based in the external described situation W meanings based in the internal (evaluative/perceptual/cognitive) described situation; (II) Meanings based in the external or internal described situation W meanings based in the textual and metalinguistic situation; (III) Meanings tend to become increasingly based in the speaker’s subjective belief state/attitude toward the proposition.
Current trends in diachronic semantics and pragmatics
3
Most recent discussions in the fields of diachronic semantics and pragmatics revolve, as far as we can tell, around refining this paradigm. In particular, alternative proposals center on four key issues, which we shall elaborate on in the remainder of this introduction, highlighting the contribution made by the present volume. 1.1. What pragmatic entities are involved? The first issue concerns the nature and likely sequence of the pragmatic entities involved in the model, for example, the role of generalized versus particularized conversational implicatures (PCIs) in language change and their relationship to the issues of propagation and actualization. In fact, reflection on diachronic models for change has led to renewed discussion of the characteristics of key pragmatic notions, such as that of implicature. Thus, the three-stage model outlined in the IITSC, whereby semantic change would proceed from PCI via GCI to coded meaning, has been criticized, for instance, by Hansen and Waltereit (2006: 248), in whose opinion: in formulating their model of semantic change, Levinson (1995, 2000) as well as Traugott and Dasher (2002) redefine the notion of GCIs. From being a purely pragmatic phenomenon, GCIs, as understood in connection with the macro-sequence under review, become something more akin to the phenomena of either propagation or actualization of specific linguistic changes.
The authors present an alternative proposal, grounded in the assumption that PCIs are ‘‘prototypically in the communicative foreground of messages, whereas generalized conversational implicatures are prototypically backgrounded’’ (Hansen and Waltereit, 2006: 235). The sequence PCI-GCI-coded meaning is argued to be rare, most cases of semantic change being either instances of a PCI semanticizing directly (PCI-coded meaning) or of a GCI semanticizing, but only after having being foregrounded as a PCI (GCI-PCI-coded meaning). Indeed, the necessity of an intermediate GCI stage would seem to simply preclude semantic change based on metaphor, for the latter relies, in the Gricean view, on ostensive flouting of the quality maxim, a maxim which according to Levinson (2000: 74) cannot generate GCIs at all. In those cases where a PCI does turn into a GCI, it is, in fact, likely that it will not become fully semanticized (PCICGI-*coded meaning). In other words, the role of the different types of implicature in meaning change is a good deal more nuanced than the IITSC model would suggest. The issue of the nature of the pragmatic entities involved in change is addressed from a different angle by Schwenter and Waltereit (forthcoming). These authors provide an outline of the evolution of the additive particle too and some of its counterparts in Spanish and German, to show how counter-argumentative uses of the particle in novel discourse contexts can override the additive presupposition normally associated with such particles. Accommodating that
4
Maj-Britt Mosegaard Hansen and Jacqueline Visconti
presupposition becomes too costly and leads to reanalyzing the too element as a new formfunction pairing. The initial diachronic examples in bridging contexts are plausibly interpretable as conveying the additive meaning of too, but the adversative properties of the dialogal discourse context appear to have led hearers to understand too as expressing a new, rhetorically strategic meaning with strong counter-argumentative force. In a similar vein, Eckardt (this volume) proposes that language change may be triggered, not just by implicatures, but also by unwarranted presuppositions. She shows that, antedating the emergence of the contemporary senses of certain words, one finds uses of those words in which their presuppositions are violated, in the sense that they contradict common world knowledge. Eckardt refers to such instances as ‘‘little pragmatic accidents’’. Many such ‘‘accidents’’ may best be repaired by a listener who hypothesizes that the problematic item is in fact being used in a new sense. According to this author, such utterances create a pragmatic overload: under a conservative interpretation, they will trigger presuppositions that the listener cannot accommodate because they do not make sense. Thus, the listener can either (a) be uncharitable and refuse to interpret the utterance at all, (b) face the pragmatic overload and attempt to reconceptualize the world such that the presupposition is consistent, or (c) hypothesize a new meaning for parts of the utterance, including the item that gave rise to the problematic presupposition. Proposals such as these raise the further issue of the role and scope of the Uniformitarian Principle in diachronic semantics and pragmatics. Within the field of linguistics, the Uniformitarian Principle consists in the assumption that the processes underlying instances of language change in the past were essentially the same as those that can be seen to operate today (e.g., Labov, 1994: 21). Among the contributors to the present volume, Eckardt explicitly argues that a virtue of her analysis is its conformity with the Uniformitarian Principle, and many of the remaining contributors tacitly rely on that principle in their analyses. However, while the principle can probably be uncontroversially applied in the study of phonological change, for instance, given that the physiological properties of speech have not changed, it becomes more problematic in the analysis of meaning change; for, as a number of scholars have argued, speech acts and events, norms of politeness, principles of text structure, and conversational routines are by no means directly comparable neither across contemporary cultures nor across different historical stages of a given culture or society (e.g., Wierzbicka, 1991, 2006; Jacobs and Jucker, 1995; Arnovick, 1999; Scollon and Scollon, 2001). To the extent that this is true, how can we be justified in assuming that patterns of inference, and hence the fundamental pragmatic entities underlying them, such as presuppositions and implicatures, were similar to those that we take to be operative in contemporary Western discourse? In those cases where the nature of both an older ‘‘source’’ meaning and a newer ‘‘target’’ meaning of a given linguistic item or construction is well established, it seems legitimate to suppose that the motivations for and mechanisms of extension/change which would most plausibly get us from source to target according to contemporary patterns of inference are likely to have been likewise instrumental in
Current trends in diachronic semantics and pragmatics
5
bringing about the change in the past.2 As soon as reconstruction is involved to a more significant extent, we are on shakier ground, however. For this reason, as current theorizing about diachronic semantics and pragmatics, as exemplified by the contributions to this volume, develops and is refined, there can be little doubt that the field would benefit from being informed to a greater extent by insights from ‘‘pragmaphilology’’ (Jacobs and Jucker, 1995) or ‘‘historical discourse analysis proper’’ (Brinton, 2001), that is, the synchronic study of discourse–pragmatic structures and functions and their corresponding means of expression in older texts. At present, however, there appears to be relatively little overlap among practitioners of the two disciplines. 1.2. What are the respective roles of the speaker and the hearer? A second major issue in diachronic semantics and pragmatics concerns the redefinition of the respective roles of speakers/writers and addressees/readers in the process of innovation. In particular, a set of proposals criticizes the speaker-based nature of the IITSC approach. As noted by Traugott (1999: 95): Although it recognizes the importance of guiding addresses to an interpretation [ . . . ] nevertheless the assumption of IITSC is that the speaker/writer does most of the work of innovation, not the hearer/reader. The idea is that the speaker/writer tries out a new use exploiting available implicatures. If the innovative use succeeds, the hearer/reader will interpret the intention correctly, and possibly experiment in similar ways in producing speech/writing. But rarely does the act of interpretation itself lead directly to innovation.
Although this is to some extent a classic chicken-and-egg problem, Traugott’s assumption has been challenged by several scholars, among whom the authors of the two parallel studies already mentioned above, both proposing a model of semantic change based on the diachronically less used notion of presupposition. In the view of both Schwenter and Waltereit (forthcoming) and Eckardt (this volume), hearers who are unable or unwilling to accommodate presuppositions assume a novel interpretation of a former presupposition trigger and eventually pass this new interpretation on to other people, thereby changing the language. According to Schwenter and Waltereit, a hearer who is confronted with an utterance and assigns an interpretation to it that deviates from that utterance’s literal meaning has, in principle, two options: (i) assume that regular pragmatic operations, such as
2
Even this cannot be taken for granted, however: when scholars rely on their semantic–pragmatic intuitions, competing mechanisms may sometimes be invoked with equal probability to account for one and the same instance of change.
6
Maj-Britt Mosegaard Hansen and Jacqueline Visconti
conversational inference and the accommodation of presuppositions, mediate between the literal meaning and the chosen nonliteral interpretation; (ii) assume a novel conventional meaning for some element of that utterance. In that case, fewer pragmatic computations, and/or less costly ones, may be required, since there is no need anymore for them to mediate between the chosen interpretation and the traditional (previous) conventional meaning. If a hearer chooses (ii) and uses the novel form-meaning pairing in his/her own discourse, the language will have changed. Assuming that interpretive efforts by the listener are at the heart of ‘‘Avoid-PragmaticOverload’’-induced changes, Eckardt (this volume) moreover suggests a hearer-driven conception of the trend toward subjectification. Earlier approaches, she argues, leave open the question whether the subjective element is pressed into an utterance as a kind of Sprachnot phenomenon by the speaker or enters the meaning of words by an interpretive effort of the listener. Eckardt suggests that, in at least some cases, the subjective element enters the language via the interpretive efforts of the listener, not because listeners are constantly searching to see the speaker’s soul through his utterances, but because in the case of a truly senseless utterance the charitable listener will try to make at least some sense of that utterance. It is much more implausible to perceive subjectification as a speaker-driven process. Uttering incoherent sentences and hoping that the listener will grab your message does not seem a rational communicative strategy (Eckardt, this volume).
Hansen (2008: 76) likewise argues that the reanalyses performed by hearers may be unintended by speakers, and that they may even, on some occasions, result from clear misunderstandings on the hearer’s part. In particular, theories that attribute a central role to the metonymical processes cannot afford to ignore this possibility, in so far as literal and metonymical meanings will often be mutually compatible. To take a simple example, if the change from Latin TESTIMONIUM (testimony) W French te´moin (witness) came about as suggested by Koch (2004: 16f ), namely via the (necessary) metonymical link between the existence of testimony and that of a witness, in contexts where utterances such as (1) would be produced, then it is unlikely that it would have been speaker driven: (1)
[Judge:] Audiamus testimonium proximum! ‘Let’s hear the next piece of testimony!’
Even more obviously, while the change responsible for the two currently competing meanings of the Danish noun bjørnetjeneste (literally ‘bear favor’), ‘a favor that has unintended negative results’ W ‘a really big favor’, can be attributed to the non-transparency of that noun, and hence its potential ambiguity in contexts such as (2), it is highly implausible that speakers familiar with the original, negatively loaded, meaning could be held responsible for the rise of the innovative, positively loaded, interpretation. In this case, there can be little doubt that the process must have
Current trends in diachronic semantics and pragmatics
7
been driven by hearers who were unfamiliar with the intended meaning and who were therefore forced to make conjectures: (2) Peter overtalte chefen til at give mig opgaven. Det var sørme en bjørnetjeneste! Literally: ‘Peter persuaded the boss to give me the task. That was really a bear-favor!’
1.3. In what types of context does semantic–pragmatic change take place? The third rapidly evolving area of discussion concerns the characterization of the contexts in which inferencing takes place. Invited inferencing and context-induced reinterpretation both assume pragmatic polysemy and ambiguity. In this respect a set of proposals have been advanced in the literature about such ambiguous contexts. Heine (2002: 86), for instance, proposes a series of four stages in semasiological change, and sees the pivotal stage as constituted by what he refers to as ‘‘bridging contexts’’, that is, contexts where the use of a given expression allows, in addition to its conventional meaning, an inference to an innovative meaning (see also Evans and Wilkins, 2000: 549; Enfield, 2005: 318).3 Putting more emphasis on structural constraints, Diewald (2002, 2006) proposes that ambiguity and structural factors accumulate in one context, which she calls a ‘‘critical context’’. An innovative element in the reflection on this topic has been the increasing awareness of the importance of taking interactional factors into account when defining the contexts for change, for instance, dialogic and contesting contexts evoking multiple viewpoints, turn-taking, and other interactional moves. The refinement of context characterization is a main thread throughout the papers in this volume. Such a refinement is operated at multiple levels: from linguistic features (co-text) to cognitive and/or interactional properties (context) to features of specific genres influencing semantic change (discourse tradition). At the linguistic level, Evans’, Jensen’s, and Romera’s papers offer fine-grained analyses of the formal and functional co-textual variables to which changes can be attributed. Thus, a detailed reconstruction of the change from aspect/mood marker to discourse particle in Morovo is related by Evans to the morphosyntactic and semantic characteristics of the construction. In Jensen’s paper, the semantics of predicates is shown to interact with properties of specific slots in a topological model of Danish sentences, thereby affecting the development of the affirmative markers godt and vel; while in Romera’s contribution, a set of formal and functional variables are shown to contribute to the evolution of Spanish es que.
3
Whereas Heine (2002) assumes that the innovative (target) meaning is the intended one, Hansen (2008: 63) proposes that it is still backgrounded at this stage, and only foregrounded when the subsequent ‘‘switch-context’’ stage is reached.
8
Maj-Britt Mosegaard Hansen and Jacqueline Visconti
The second level, that of cognition and interaction, which is not always given prominence in diachronic semantics and pragmatics studies, is central to many papers in the present volume: Hansen draws attention to what she calls ‘‘Janus-faced contexts’’, that is, contexts in which the cognitive contents of a negated clause are oriented simultaneously to previous and to upcoming discourse, and proposes that such contexts are crucial to the grammaticalization of bipartite sentence negation in French. Paradigmatic associations (also treated by Diewald et al. and Jensen) are argued to have predictive value by Estelle`s, who shows how paradigmatic pressures can lead to change. Bazzanella and Miecznikowski analyze how Italian allora has evolved from expressing referential (temporal) meaning to fulfilling a range of inferential functions, and argue that the driving forces in these changes should be sought among properties of spoken language in interaction (e.g., planning, recipient designed, discourse structuring) as well as among properties of co-constructed argumentational discourse (e.g., recurrent schemas of reasoning; the need to attribute conclusions and premises to participants and to monitor the degree of sharedness/ expectedness of standpoints at any moment of their negotiation; the direct relevance of dialogical reasoning to decisions about future actions). Similarly, Detges and Waltereit argue that discourse markers arise as the result of argumentational procedures in the negotiation in discourse, marking, for example, a change of activity or disagreement, while the polyphonic component in modal particles arises as the conventionalization of disputing the validity of a given proposition in dialogic exchanges. Finally, Beeching highlights the crucial role of concessive contexts in pragmaticalization processes such as the evolution of hedging and boosting particles in a variety of languages. At a more general level, Stathi’s study of the development of the German verb geho¨ren (literally ‘belong to’) shows how discourse traditions (in this case, administrative and judicial texts) may influence change, an aspect previously discussed by Pons Borderı´a (2006). 1.4. What is the precise nature of and relationship among the observed tendencies of semantic–pragmatic change? The fourth issue revolves around both refining the clines (i) to (iv) identified above and understanding the relationship between such tendencies: for instance, the relationship between grammaticalization, scope, and subjectification (Company Company, 2006a,b; Traugott, forthcoming), or the issue of whether to define the evolution of pragmatic markers as an instance of grammaticalization (Brinton and Traugott, 2005; Diewald et al., this volume) or of pragmaticalization (Erman and Kotsinas, 1993; Dostie, 2004; Hansen, 2008: Chapter 3). With respect to the latter, those who argue that pragmatic markers are grammaticalized focus on the fact that the evolution of pragmatic markers will typically feature the decategorialization of the source item, some degree of phonological reduction of that source item, as well as subjectification and increased ‘‘procedurality’’ of its content.
Current trends in diachronic semantics and pragmatics
9
Those scholars who prefer to describe pragmatic markers as ‘‘pragmaticalized’’, on the other hand, do so on the basis of Lehmann’s (1985) six classic parameters of grammaticalization, viz. phonological and/or semantic attrition, paradigmatization, obligatorification, scope reduction, syntagmatic coalescence, and syntagmatic fixation. According to Lehmann’s model, grammaticalized items will exhibit a high degree of several of these characteristics. As noted by Waltereit (2002: 1005), Eckardt (2003: 42), and Hansen (2008: 57f ), pragmatic markers as a class tend not to fulfill Lehmann’s criteria to any great extent. Rather, they typically exhibit scope increase, greater syntactic freedom, optionality, and strengthening of their pragmatic import. The former problem has to do with whether subjectification is necessarily part of, and unique to, the grammaticalization process, which, if true, raises questions about the scope of grammaticalized items. According to Traugott (1995) and Brinton and Traugott (2005), subjectification is at the very least a strong tendency in grammaticalization, particularly in its early stages (Traugott, 1995: 47), but it is nevertheless an independent process (Brinton and Traugott, 2005: 109; Traugott, forthcoming). Company Company (2006a) argues that subjectification is not just a semantic–pragmatic phenomenon, but actually constitutes a specific type of syntactic change as well, namely one that is characterized by scope increase and syntactic isolation, and she conceives of subjectification as a subtype of grammaticalization. It seems to us that what this author calls subjectification would largely be equivalent to what others describe as pragmaticalization, were it not for the fact that she deems the syntactic changes mentioned criterial, as opposed to just typical, of subjectification. However, the existence of items like the socalled modal particles that are a salient feature of the Continental Germanic languages appears to provide a strong argument against such a view, for while modal particles are indubitably subjectified as compared to their source items and also exhibit scope increase, they are nevertheless syntactically highly constrained. This debate raises the issue of the exact understanding of what is implied by notions such as grammaticalization and pragmaticalization: are these labels for specific processes of change, in which case they have independent theoretical status, or are they largely just convenient shorthands for the cumulative results of sets of frequently converging, but essentially independent, changes that linguistic items and constructions can undergo? If the former, we would expect there to be a rather strict separation between grammaticalized and pragmaticalized items. If the latter, we should not be surprised to find cases where either or both label(s) might seem appropriate. This would appear to be the case, for instance, with the Germanic modal particles mentioned above, which are characterized by scope increase, optionality, and pragmatic strengthening, but also – unlike discourse markers, for instance – by paradigmatization and syntagmatic fixation. The issues surrounding grammaticalization, pragmaticalization, and processes, such as subjectification which may be prominently associated with either, are highlighted in many of the papers in the present volume, which analyze the evolution of procedural meanings of various kinds. Thus, several papers feature different types of pragmatic markers as their object of study
10
Maj-Britt Mosegaard Hansen and Jacqueline Visconti
(Bazzanella and Miecznikowski, Beeching, Detges and Waltereit, Eckardt, Estelle`s, Evans, Gehweiler, Jensen, Romera), while others are concerned with items and constructions expressing modality (Stathi), evidentiality (Diewald et al., Squartini), negation (Hansen), and relational meanings (Adler and Asnes). Saliently, Diewald at al. refine the grammaticalization model as the reinterpretation and abstraction of a relational semantic template from (i) referential to (ii) textintegrative/connective to (iii) indexical-grammatical function, exemplified by the evolution of evidentials and modal particles in German, while Detges and Waltereit’s tackle the thorny issue of the categorization of discourse markers versus modal particles, which the authors claim arise from different mechanisms of change. Their paper proposes a fine-tuning of the subjectification model by showing how and why different types of procedural meanings arise. 1.5. Concluding remarks We suggest that, in a number of ways, the present volume constitutes an important contribution to our current understanding of historical semantics and pragmatics: First, several papers in this collection revisit, in a diachronic perspective, key theoretical notions that are typically confined to the synchronic perspective, such as presuppositions (Eckardt), paradigms (Estelle`s), word order (Jensen), and discourse status (Hansen). This allows the authors to test and refine current models of semantic change and to provide innovative accounts of causes and motivations for linguistic changes. Second, the semantic domains covered by the case studies are ones that are generally considered central (spatiality, temporality, negation, modality, evidentiality, subjectivity, scalarity, intensification, and concession). Third, the volume commends itself not only by virtue of the variety of approaches to meaning that are represented here, from prototype theory (Bazzanella and Miecznikowski) to monosemy (Adler and Asnes), but also very much by the range of data adduced from languages other than English (several Romance languages, German, Danish, and Oceanic languages).
2. SUMMARIES
OF THE
INDIVIDUAL CONTRIBUTIONS
The papers collected in this volume originate in a workshop on Diachronic Semantics and Pragmatics, organized by the volume editors at the 18th International Conference on Historical Linguistics (Montreal, August 2007). In her article, ‘‘Avoid Pragmatic Overload,’’ Regine Eckardt suggests that the role of pragmatics in language change might not be restricted to implicature, but that presupposition (failure) is equally a driving force in meaning change. It is well known that utterances may carry
Current trends in diachronic semantics and pragmatics
11
presuppositions which, if new to the hearer, will have to be accommodated. This is unproblematic in all those cases where the accommodated information is plausible and uncontroversial. Problems arise, however, when the speaker relies on presuppositions which are implausible, controversial, or hard to reconcile with other pieces of knowledge. Eckardt proposes that when faced with the option of accommodating the impossible, hearers may instead prefer to reanalyze the meaning of parts of the utterance. Thus, hearers avoid pragmatic overload (problematic presuppositions) and hypothesize new meanings instead. The proposed analysis is supported by several attested cases of semantic change to which it can be fruitfully applied. Among others, the author discusses the reanalysis of German intensifying selbst (-self ) as a focus particle (even), and of German fast (immovably tight) as an approximative (almost), as well as the development of English even from a level adjective to a focus marker, and she takes a brief look at the case of Italian perfino which likewise develops an ‘even’-like reading from its earlier sense ‘through, to the end’ (Visconti, 2005). For each item, it can be argued that the turning point of the development is characterized by the appearance of uses where the presuppositions of the sentence, if spelled out, are tantamount to contradictory, or at least highly implausible, information. The proposal confirms the vital role of pragmatics in language change and identifies yet another type of pragmatic enrichment of utterances that has, so far, not been widely explored in diachronic linguistics. In their paper, ‘‘Diachronic Pathways and Pragmatic Strategies: Different Types of Pragmatic Particles from a Diachronic Point of View’’, Ulrich Detges and Richard Waltereit address one of the main concerns of research on discourse markers, modal particles, and related elements, namely the problem of a neat categorical delimitation between these items, by considering whether the synchronic difference between discourse markers and modal particles can be accounted for in diachronic terms. Do discourse markers and modal particles arise from different mechanisms of change? Their cases are the cognate French and Spanish particles bien, both of which originate in adverbs meaning ‘well’. French bien functions, among other things, as a modal particle. By contrast, Spanish bien is a discourse marker. It is widely accepted that discourse markers serve the purpose of coordinating the joint construal of discourse, and according to Detges and Waltereit, this is directly reflected in their diachronic evolution: thus, the Spanish DM bien is the routinized residue of negotiations about the next move in conversation. Moreover, their analysis suggests that discourse markers are a subset of a much wider range of routines that human beings have at their disposal for the coordination of joint activities. Thus, it is not surprising that the diachronic evolution of a discourse marker should follow the same pathways as do such routines. By contrast, modal particles function at the speech-act level and typically make reference to the hearer’s attitudes concerning the validity of the speech act. More broadly, the results presented in this paper imply that there are levels of generalization about semantic change below the overarching tendencies of subjectification. At the same time, they provide very specific justification for these levels of generalization, and ultimately for subjectification itself.
12
Maj-Britt Mosegaard Hansen and Jacqueline Visconti
Eva Skafte Jensen’s article, ‘‘Context Sensitive Changes: The Development of the Affirmative Markers ‘godt’(good) and ‘vel’ (well) in Danish’’, gives a detailed account of how these Danish adverbials (meaning, respectively, ‘good’ and ‘well’) came to be used as markers of affirmation. Special interest is paid to the role played by the syntactic, semantic, and pragmatic context of the lexical items in question. One influential factor is the semantics of the predicates of the sentences in which the items occur. Predicates conveying subjective meanings and meanings of modal possibility (e.g., modal verbs, verbs such as føle ‘feel’, vide ‘know’, etc.) seem to provide good conditions for the development of the affirmative function of the adverbials in question. Another, less frequently discussed, contextual factor is provided by the rules of word order (topology) in Danish. Jensen argues that the properties of specific slots and places in the topological model of a Danish sentence have direct bearing not only on the synchronic interpretation of the adverbials in question, but also on their possible diachronic reinterpretation as markers of affirmation. A third contextual factor is of a pragmatic nature, as it is argued that the developments of godt and vel may be described as ‘‘the conventionalization of a conversational implicature’’. Thus, these adverbials often occur in utterances where there might be some doubt as to the validity of the positive value of the propositional contents, thus giving rise to the implicature that ‘someone might think that the State-of-Affairs is not the case’. As a result of this, affirmative godt and vel enter into a small paradigm of polarity alongside negation and zero. The cross-linguistic implications of this account are evaluated, as some languages seem to prefer having adverbials meaning ‘good’ and ‘well’ develop into markers of affirmation or similar (e.g., Detges and Waltereit, this volume). ‘‘Procatalepsis and the Etymology of Hedging and Boosting Particles’’ by Kate Beeching sets out to explore a cognitive basis for procatalepsis, ‘a figure by which an opponent’s objections are anticipated and answered’, and to argue that the evolution of hedging/boosting particles in a number of languages may be explained by reference to it. While the role of metaphor and metonymy in language change is well documented, other classical figures such as synecdoche and procatalepsis and their relationship with cognition and semantic–pragmatic change have been less thoroughly investigated. By conceding certain arguments, speakers can strengthen their position and make it easier to defend, while at the same time exhibiting a sense of fair play. In a similar manner, a procataleptically derived particle simultaneously hedges and boosts the assertion associated with it. Beeching’s paper argues that the metonymic contiguity of terms like though with contexts relating to the negotiation of meaning and the ‘‘Cardinal Concessive frame’’ (X – statement, Xu – concession, Y – potential refutation) (Couper-Kuhlen and Thompson, 2000) so commonly invoked in everyday interaction has led them to be used in situations where X and Xu are unexpressed. Though thus retains a concessive sense, but functions simultaneously as a hedge and as a booster. What is more, this procataleptic tendency is a cross-linguistic phenomenon which affects similar adversative and concessive conjunctions and adverbs in different languages, such as French quand meme, Glasgow but and German aber. While the etymologies, the
Current trends in diachronic semantics and pragmatics
13
polysemies, and the position in the clause of these various adverbs differ, the fundamental cognitive reflex depending on the Cardinal Concessive, which is part of our everyday conceptualization of events, is very similar in all of them. This suggests that the underlying motivation for the recruitment of such usages is universal, arising from social interactional exigencies related to questions of face. The study by Carla Bazzanella and Johanna Miecznikowski, ‘‘Central/Peripheral Functions of ‘allora’ and ‘Overall Pragmatic Configuration,’’’ suggests that Italian allora (‘so’, ‘then’) has considerably expanded its functional spectrum from the 13th century up to the present day. Starting out from temporal functions (simultaneity and consecution), the following functions have been added (in order of diachronic appearance): (i) functions in the domain of causality, (ii) inferential-evidential functions, and (iii) text and interaction structuring functions. Functional expansion has been accompanied by semantic shifts, among others a deictic shift (distal W proximal), as well as by syntactic shifts (in particular, ana-/cataphorical adverb W connective at utterance margins W iconic segment-initial connective). In this development, the use of allora in hypothetical constructions – which in themselves are polyfunctional and often potentially ambiguous – seems to have played an important role. To explain allora’s semantic-functional expansion, it is useful to posit (a) a basic relational semantics of the lexeme, which is quite stable over time and (b) general semantic/cognitive principles that account for the relative proximity of certain semantic domains, and therefore the probability of a (metaphorical) shift from one domain of meaning to the other. However, (a) and (b) merely delineate potential paths of development; in the explanation of change, they have to be integrated with specific hypotheses about the driving forces of change. Bazzanella and Miecznikowski strongly emphasize the importance of pragmatic factors as driving forces, that is, functional pressures arising in recurrent situations of use. In the case of allora, strongly argumentational dialogue types seem to have played a key role. In the analysis offered, the shifts mentioned above are related to both general properties of spoken language in interaction and more specific properties of co-construed argumentational discourse. The action of these factors as driving forces can be accounted for if one assumes that the semantic and functional potential of a linguistic item corresponds to a prototypically organized set of features which, in language use, is actualized as an ‘overall pragmatic configuration’, in which some potential features become more relevant than others, in congruence with contextual parameters, and are thus strengthened by recurrent use in certain constructions and situation types. Unlike routinization (Hopper, 1987) and context (Hopper and Traugott, 1993 [2003]), both traditionally regarded as triggers of grammaticalization, a factor such as paradigmatic relations has mainly been seen as responsible for the spread of change, not as the locus where changes take place. Against this background, Maria Estelles’ article on ‘‘The Importance of Paradigms in Grammaticalization: Spanish Digressive Markers ‘por cierto’ and ‘a propo´sito’’’ highlights their role as a possible cause of grammaticalization processes. The importance of paradigms is illustrated by tracing back the history of the two most frequent markers of digression in Spanish,
14
Maj-Britt Mosegaard Hansen and Jacqueline Visconti
namely por cierto and a propo´sito (by the way). These particles, despite rather different origins and evolutions, seem, in the course of the 19th century, to become integrated into a paradigm of digression. Since then, the use of por cierto, formerly restricted to the intra-sentential level, has been extended to the textual level, already reached by a propo´sito. In turn, a propo´sito came to be used in appositions, an environment that was previously specific to por cierto. Magdalena Romera’s ‘‘The Multiple Origin of ‘es que’ in Modern Spanish: Diachronic Evidence’’ considers how an expression such as es que (it is that), originally part of a structure that expresses existential meaning, ends up being used by itself as a functional unit to introduce elaboration–reinterpretation values. Her proposal is that constructions with es que initially express existential meaning and a more elaborated clarification of the content of a previous segment, but later the elaboration allows for a more subjective interpretation in terms of the speaker’s interpretation, which in turn could be understood as an explanation for what was just said. At the same time, es que constructions lose their subject due to a process of loss of referentiality in the elements placed in that position. Initially, subjects are highly referential and can be related anaphorically to the previous content, but later on they are simply elements that anticipate the focalized content expressed in the subordinate clause. Chronologically, these two processes go together. From Early Spanish (1200–1300) up to the 16th century, es que constructions are existential and elaborative structures. The first cases of interpretative uses are found in 1500 and generalize in the 17th century. The same can be said of the path from integrated constructions (i.e., es que constructions with a subject) to nonintegrated ones (i.e., es que constructions without a subject). No cases of subjectless constructions are found in Early Spanish and only a few cases in 1400. The first uses of es que structures as we know them in Modern Spanish appear in 1500 in monologues and at the end of that century in dialogues. The subject position is allowed to be left empty in the 16th century. Bethwyn Evans’ paper, ‘‘From Aspect/Mood Marker to Discourse Particle: Reconstructing Syntactic and Semantic Change,’’ examines the reanalysis of an aspect/mood marker as a discourse connective particle from the perspectives of both syntactic and semantic change. Evidence of the change is found in the system of subject marking in Marovo, an Oceanic language of the Solomon Islands. Marovo has preverbal markers which indicate the person and number of the subject argument and which occur primarily in only two types of constructions: negative verbal declarative clauses and verbal clauses with an initial discourse connective particle. These unusual conditions on the presence of subject marking in Marovo are shown to reflect its historical development. Through comparison of Marovo with other closely related Oceanic languages, it is demonstrated that subject marking in negative clauses is archaic, reflecting original constructions in which subject markers occurred within the verb complex alongside preverbal markers of aspect/ mood and negation. The use of subject markers with discourse connective particles reflects the same original construction, but in this case the reanalysis of an aspect/mood marker as a discourse connective particle has resulted in the subsequent extension of subject markers to use with
Current trends in diachronic semantics and pragmatics
15
discourse connective particles in general. A detailed reconstruction of the change, informed by accepted models of syntactic and semantic reanalysis, suggests that it was motivated by both the morphosyntactic and semantic characteristics of the construction. While the reanalysis appears to have been triggered by structural ambiguity resulting from a chance homophony of forms, semantic and pragmatic aspects of the construction also facilitated the change. In ‘‘The Grammaticalization Channels of Evidentials and Modal Particles in German: Integration in Textual Structures as a Common Feature’’, Gabriele Diewald, Marijana Kresic, and Elena Smirnova are concerned with the grammaticalization of German evidential constructions and modal particles. In present day German, these items serve as linguistic means for expressing different grammatical contents, and, at first sight, do not seem to have anything in common beyond being two evolving grammatical categories. The authors’ first concern in this paper is to argue that both German evidential periphrases and modal particles serve as grammatical markers. As such, they meet two criteria that are proposed as definitional of grammatical signs: (a) indexical potential and (b) paradigmatic integration. The second aim of the paper is to reconstruct a diachronic developmental path for each category. Thus, the grammaticalization processes of individual elements are summarized in a unified grammaticalization channel for each category. The third goal of the paper is to show that there are common developmental tendencies in the grammaticalization of even such different linguistic elements as evidentials and modal particles. The intention behind this is to propose that these common features may be powerful indicators of grammaticalization in general. The starting point is the assumption that, in their development, these elements follow general tendencies and clines established in grammaticalization theory, and the authors show (i) that German evidentials and modal particles develop by reinterpretation of, and abstraction from, a relational semantic template and (ii) that this development results in the indexical-grammatical interpretation of that template, which is reached via an intermediate stage of text-integrative/connective interpretation. By way of generalization, the authors assume that their model of three successive stages in grammaticalization (which describe the following semiotic-functional changes in a sign: (i) referential function W (ii) textintegrative/connective function W (iii) indexical-grammatical function) is applicable to all grammaticalization processes. Moreover, they emphasize the particular importance of the second stage in grammaticalization processes – the integration of a sign in specific text structures whereby the sign comes to serve text-connective functions. In his article, ‘‘Evidentiality, Epistemicity and Their Diachronic Connections to Nonfactuality,’’ Mario Squartini investigates the diachronic relationship between evidentiality and epistemicity in the pragmatic and semantic evolution of some Romance forms, including seemverbs, hearsay markers such as the American Spanish dizque ‘allegedly, supposedly’, and inflectional verb forms having epistemic meanings or being used as evidential strategies (the Romance synthetic futures and conditionals). Despite their diverse origins, all these forms evolve in similar diachronic directions, demonstrating the crucial role of nonfactuality in the evolution of
16
Maj-Britt Mosegaard Hansen and Jacqueline Visconti
the evidential meaning connected to hearsay and reportive markers. Nonfactuality, on the other hand, turns out to be negatively correlated with the diachronic evolution of conjectural and inferential markers, a fact which raises empirical questions with respect to the much-debated interpretation of inferentiality as an intermediate area in the crucial boundary between epistemicity and evidentiality. Maj-Britt Mosegaard Hansen’s paper on ‘‘The Grammaticalization of Negative Reinforcers in Old and Middle French: A Discourse-Functional Approach’’ presents evidence that key stages in the diachronic evolution of clausal negation in French should be understood as governed by discourse-functional constraints on the flow of information. Specifically, concerning the synchronic properties of Old and Middle French negation, she argues that clauses negated by ne . . . mie/pas were constrained to be discourse-old, as defined by Birner (2006), and that, while the proposition expressed by such clauses need not be believed, it should be such that the speaker could assume that it was either already activated in the short-term memory of the hearer or accessible to activation based on other propositions thus activated. This analysis presents the advantage of being compatible with what is known about the uses of different forms of negation in a number of contemporary Romance vernaculars where variation is still maintained between simple preverbal negators and reinforced expressions which in several cases are etymologically identical to the French forms. The author suggests, further, a diachronic scenario capable of explaining the subsequent unmarking of reinforced negation in French, her proposal being that so-called Janus-faced contexts, that is, contexts that are at one and the same time backwards and forwards oriented in terms of the flow of information in discourse, constituted the key bridging contexts that allowed for the reanalysis of the reinforced negators. The advantages of the proposed scenario are that it not only relies on precisely those discourse-functional constraints that were argued to govern the use of negative reinforcers at the stage where they were still conceptually and textually marked, but it is also more precise than existing pragmatically based explanations. Silvia Adler and Maria Asnes, in their ‘‘A Roots Journey of a French Preposition,’’ propose a diachronic investigation of the French preposition jusqu’a` (meaning ‘until’, ‘up to’, ‘to’). Their study reveals that all the present usages of jusqu’a` – spatial, temporal, scalar, and quantificational – already coexist at the early stages of French, which supports the hypothesis of the monosemicity of this preposition. Thus, all the possible readings of PPs headed by jusqu’a` share one semantic primitive which has to do with the notions of a path and of a culminating point (representing the limit of the path). It is the nature of the limit provided by the context which accounts for the different readings of jusqu’a`. This suggests, in other words, that the spatiotemporal value cannot be considered as the core meaning of this preposition: in the case of jusqu’a`, there is no real evolution from spatial to nonspatial, from concrete to abstract, but rather one core sense, in itself abstract, which is applicable to various domains, such as space, time, and scalarity. All of these conserve the idea of an axis, a continuum, and an oriented scale.
Current trends in diachronic semantics and pragmatics
17
Elke Gehweiler’s ‘‘The Grammaticalization of Privative Adjectives: The Case of ‘mere’’’ argues that the present-day English adjectival intensifier/downtoner mere evolved from a privative adjective with the meaning ‘pure, unmixed’ through a process of grammaticalization and subjectification. The paper first discusses the synchronic status of mere, showing that mere is a peripheral member of the word class adjective and is restricted to only a very limited number of patterns in PDE. In the second part of the paper, the diachronic development of mere from privative adjective to downtoner is discussed on the basis of diachronic corpus data. Most importantly, it is argued that the ambiguity of mere in attributive position in certain uses triggered a reanalysis of mere as intensifier, which was followed by the lexicalization of its new meaning. The paper suggests that the case of mere is not unique and that other privative adjectives have developed more grammatical and more subjective meanings in a similar way. Katerina Stathi’s ‘‘The Origin of Semantic Change in Discourse Tradition: A Case Study’’ shows how the textual context or discourse tradition in which semantic change of a lexical item originates may be reflected in the meaning of that item. The German verb geho¨ren (literally ‘belong to’) expresses necessity and obligation in a construction with the passive participle. Diachronic evidence reveals that this meaning arose in contexts of law and administration via pragmatic inference. A synchronic corpus study shows that a significant proportion of the participles in this construction refers to ‘‘negative’’ actions (which can be subsumed under the notion of punishment) on the patient. It is claimed that the dominant meaning of the participles reflects the original context in which geho¨ren developed the meaning of necessity and obligation. This is described as an instance of persistence.
REFERENCES Arnovick, L. K. (1999). Diachronic pragmatics. Amsterdam: John Benjamins. Birner, B. J. (2006). ‘‘Semantic and pragmatic contributions to information status’’, in M.-B. M. Hansen and K. Turner (eds.), Explorations in the semantics/pragmatics interface. Special issue of Acta Lingvistica Hafniensia, 38: 14–32. Brinton, L. J. (2001). ‘‘Historical discourse analysis’’, in D. Schiffrin, D. Tannen and H. E. Hamilton (eds.), Handbook of discourse analysis. Oxford: Blackwell, 138–160. Brinton, L. J. and E. T. Traugott (2005). Lexicalization and language change, (research surveys in linguistics). Cambridge: Cambridge University Press. Company Company, C. (2006). ‘‘Subjectification of verbs into discourse markers: Semantic–pragmatic change only?’’. Belgian Journal of Linguistics 20: 97–121. Company Company, C. (2006). ‘‘Zero in syntax, ten in pragmatics: Subjectification as syntactic cancellation’’, in A. Athanasiadou, C. Canakis and B. Cornillie (eds.), Subjectification: Various paths to subjectivity, (cognitive linguistics research 31). Berlin: Mouton de Gruyter, 375–397. Couper-Kuhlen, E. and S. A. Thompson (2000). ‘‘Concessive patterns in conversation’’, in E. Couper-Kuhlen and B. Kortmann (eds.), Cause–condition–concession–contrast: Cognitive and discourse perspectives. Berlin: Mouton de Gruyter, 381–410.
18
Maj-Britt Mosegaard Hansen and Jacqueline Visconti
Diewald, G. (2002). ‘‘A model for relevant types of contexts in grammaticalization’’, in I. Wischer and G. Diewald (eds.), New reflections on grammaticalization. Amsterdam: John Benjamins, 103–120. Diewald, G. (2006). ‘‘Context types in grammaticalization as constructions’’, Constructions, SV1-9. http:// www.constructions-online.de/ Dostie, G. (2004). Pragmaticalisation et marqueurs discursifs. Analyse se´mantique et traitement lexicologique. Brussels: De Boeck/Duculot. Eckardt, R. (2003). The structure of change. Meaning change under reanalysis. Habilitation thesis, Berlin: Humboldt Universita¨t (Revised version published in 2006 by Oxford University Press). Enfield, N. J. (2005). ‘‘Micro- and macro-dimensions in linguistic systems’’, in S. Marmaridou, K. Nikiforidou and E. Antonopoulou (eds.), Reviewing linguistic thought: Converging trends for the 21st century (Trends in linguistics. Studies and monographs, Vol. 161). Berlin: Mouton de Gruyter, 313–325. Erman, B. and U.-B. Kotsinas (1993). ‘‘Pragmaticalization: The case of ba’ and you know’’. Studier i modern spra˚kvetenskap 10: 76–93. Evans, N. and D. Wilkins (2000). ‘‘In the mind’s ear: The semantic extensions of perception verbs in Australian languages’’. Language 76: 546–592. Grice, H. P. (1989[1975]). ‘‘Logic and conversation’’, in his studies in the way of words. Cambridge: Cambridge University Press, 22–40. Hansen, M.-B. M. (2008), Particles at the semantics/pragmatics interface: Synchronic and diachronic issues. A study with special reference to the French phasal adverbs, (Current research in the semantics– pragmatics interface, Vol. 19). Oxford/Bingley: Elsevier/Emerald. Hansen, M.-B. M. and R. Waltereit (2006). ‘‘GCI theory and language change’’, in M.-B. M. Hansen and K. Turner (eds.), Explorations in the semantics/pragmatics interface. Special issue of Acta Lingvistica Hafniensia, 38: 235–268. Heine, B. (2002). ‘‘On the role of context in grammaticalization’’, in I. Wischer and G. Diewald (eds.), New reflections on grammaticalization. Amsterdam: John Benjamins, 103–120. Hopper, P. J. (1987). ‘‘Emergent grammar’’. Berkeley Linguistics Society 13: 139–157. Hopper, P. J. and E. C. Traugott (2003). Grammaticalization. 2nd ed. Cambridge: Cambridge University Press. Jacobs, A. and A. H. Jucker (1995). ‘‘The historical perspective in pragmatics’’, in A. H. Jucker (ed.), Historical pragmatics. Amsterdam: John Benjamins, 3–33. Koch, P. (2004). ‘‘Metonymy between pragmatics, reference, and diachrony’’. Metaphorik.de 7: 6–54. www.metaphorik.de Labov, W. (1994). Principles of linguistic change, internal factors. Oxford: Blackwell. Lehmann, C. (1985). ‘‘Grammaticalization: Synchronic variation and diachronic change’’. Lingua e stile XX (3): 303–318. Levinson, S. C. (1995). ‘‘Three levels of meaning’’, in F. R. Palmer (ed.), Grammar and meaning. Cambridge: Cambridge University Press, 90–115. Levinson, S. C. (2000). Presumptive meanings. The theory of generalized conversational implicature. Harvard, MA: MIT Press. Pons Borderı´a, S. (2006). ‘‘From pragmatics to semantics: esto es in formulaic expressions’’, in M.-B. M. Hansen and K. Turner (eds.), Explorations in the semantics/pragmatics interface. Special issue of Acta Lingvistica Hafniensia, 38: 180–206. Schwenter, S. A. and Waltereit, R. (forthcoming). ‘‘Presupposition accommodation and language change’’, in H. Cuyckens, K. Davidse and L. Vandelanotte (eds.), Subjectification, intersubjectification and grammaticalization, (Topics in English linguistics). Berlin: Mouton de Gruyter. Scollon, R. and S. W. Scollon (2001). Intercultural communication. 2nd ed. Oxford: Blackwell.
Current trends in diachronic semantics and pragmatics
19
Traugott, E. C. (1989). ‘‘On the rise of epistemic meanings in English: An example of subjectification in semantic change’’. Language 57: 33–65. Traugott, E. C. (1995). ‘‘Subjectification in grammaticalization’’, in D. Stein (ed.), Subjectivity and subjectivisation. Cambridge: Cambridge University Press, 31–54. Traugott, E. C. (1999). ‘‘The role of pragmatics in semantic change’’, in J. Verschueren (ed.), Pragmatics in 1998: Selected Papers from the 6th International Pragmatics Conference. Vol. 2. Antwerp: International Pragmatics Association, 93–102. Traugott, E. C. (forthcoming). ‘‘Revisiting subjectification and intersubjectification’’, in H. Cuyckens, K. Davidse and L. Vandelanotte (eds.), Subjectification, intersubjectification and grammaticalization, (Topics in English linguistics). Berlin: Mouton de Gruyter. Traugott, E. C. and R. B. Dasher (2002). Regularity in semantic change. Cambridge: Cambridge University Press. Visconti, J. (2005). ‘‘On the origins of scalar particles in Italian’’, in M.-B. M. Hansen and C. Rossari (eds.), The evolution of pragmatic markers. Special issue of Journal of Historical Pragmatics, 6(2): 237–261. Waltereit, R. (2002). ‘‘Imperatives, interruption in conversation, and the rise of discourse markers: A study of Italian guarda’’. Linguistics 40 (5): 987–1010. Wierzbicka, A. (1991). Cross-cultural pragmatics. Berlin: Mouton de Gruyter. Wierzbicka, A. (2006). English. Meaning and culture. Oxford: Oxford University Press.
Current Trends in Diachronic Semantics and Pragmatics Edited by Maj-Britt Mosegaard Hansen and Jacqueline Visconti r 2009 by Emerald Group Publishing Limited. All rights reserved.
2 APO: AVOID PRAGMATIC OVERLOAD Regine Eckardt
1. CHANGE
WITHOUT
MYSTERIES
Much recent work in historical linguistics, notably historical pragmatics, is focused on the mysterious moment where a new item or construction sees the day of the light. Theoretically speaking, such a situation should have the following characteristics: (1) There is a speaker who makes an utterance: u u is still part of the old language stage Lold The hearer understands: uu uu is part of the new stage Lnew If we really want to capture moments of change, we must assume that the hearer was not already competent in the new language stage Lnew before s/he understood uu. The innovation is constituted by the hearer parsing the utterance and deriving a meaning in a way that differs from what the speaker had in mind with her utterance u. There are several proposals in the literature about how situations of this kind can arise. One very simple scenario really avoids all difficulties by claiming that new language stages typically come about by innovative acts by the speaker. The speaker can decide to use language in innovative ways, and to the extent that the hearer can make sense of an innovative utterance, and adopts the suggested underlying pattern, the hearer confirms and adopts the new language stage. This view is already inherent in traditional work in language history (von der Gabelentz, 1891; Paul, 1920) and it is moreover extremely plausible, because we can observe innovative utterances on a daily basis. Yet, when we think about the origin of the more routine parts of language – aspect forms, particles, tenses – it is unlikely that all these have come about by witty remarks of creative speakers.
22
Regine Eckardt
According to another scenario, innovations can enter a language because hearers are incompetent, simply misunderstanding the intended linguistic structure of the speaker’s utterance. Indeed, we know that an increased rate of potential misunderstandings, for example, in large L2 communities, may lead to increased speed in language change but there are interesting cases in the histories of languages where no such driving force would be known. In recent work, Traugott and Dasher (2002) have devised another detailed scenario which illustrates the above type of situation. Traugott in fact was the first to point out that pragmatic processes are a driving force in language change (Ko¨nig and Traugott, 1988; Traugott, 1988), and this is explored in her theory of implicature-based language change. An utterance can mean more than its literal meaning (implicatures). Implicated information can turn into lexically denoted information (generalized invited inferences, GIINs). Due to this reinterpretation (by the hearer), items and constructions can change in meaning. Again, the appeal of the analysis lies in the fact that it rests on pragmatic processes that can be witnessed all over the place in contemporary communication. Hence, the GIIN theory of language change adheres to the uniformitarian principle. In the present paper, I want to draw attention to yet another pragmatic factor that can lead to utterance-comprehension mismatches of the type in (1), namely the presuppositions of utterance u. A sentence u presupposes further information f if u only makes sense at all if f is known. Definite noun phrases like ‘the king of France’ are typical textbook examples. Most sentences that contain the NP ‘the king of France’ will only make sense – whether they be true or false – if there is such a person in the first place. If a speaker utters a sentence with presupposed information, s/he relies on shared knowledge between speaker and hearer. How can presuppositions give rise to language change? Sometimes, a speaker will utter a sentence u, which presupposes information f that the hearer actually did not know before. In this case, the hearer will frequently just tacitly adopt f as another piece of new information that the speaker seems to believe (or else, the speaker would not have uttered u). If the hearer feels that f is totally unwarranted, s/he can object (‘hey, listen, there is no king of France’). In the present paper, I will investigate a further kind of semantic accident that presuppositions can cause, one that, to my knowledge, has not received attention in either semantics/pragmatics or historical linguistics. We find linguistic exchange where an utterance u presupposes information f that is ‘‘hard to believe’’ not in the sense that it would be a proposition with clear but dubitable content. Sometimes, presuppositions are ‘‘hard to believe’’ in that it is unclear what the presupposed facts that would license an utterance could look like at all. Hearers (or readers) of such utterances will diagnose that (i) either the speaker believes facts about the world that are unclear and dubious or that (ii) the speaker might have used words or phrases in a sense that were formerly unknown to the hearer. If the hearer pursues hypothesis (ii), s/he may come to interpret the utterance in some innovative way uu that defines a new language (micro) stage Lnew even though the speaker firmly believed that he/she was making an utterance u in the conservative Lold. From the speaker’s perspective, all the hearer would have had
Avoid pragmatic overload
23
to do is adopt-and-believe some presuppositions (we will use the official term accommodation later). From the hearer’s perspective, it was harder to accommodate the presupposed information than to believe that the utterance was really something new. The utterance created too much pragmatic overload. This is the abstract backbone of the proposal. I will discuss four example cases where I believe that this proposal is better in line with the attested uses at certain phases of semantic change than either one of the accounts that I listed at the beginning. I can not exclude that the real change came about in ways different from those that I will devise here. However, the present proposal is an attempt to make sense of data in phases of change, which are hard to reconcile with other analyses of change, notably those listed at the beginning. In the next sections, I will introduce the examples and will list the open questions that are posed. I will then offer a more detailed introduction of presuppositions and presupposition accommodation, the core concepts of my proposal. We will then see how change arises from pragmatic overload in each of the examples at stake.
2. IMPLICATURES
ARE NOT
ENOUGH
In this paper, I will be concerned with the following four items. All of them passed at least once from one meaning to a subsequent new meaning, as listed here. German English German Italian
fast even selbst perfino
(1) (1) (1) (1)
‘immovably, tight’W(2) ‘very much’W(3) ‘almost’ ‘flat, smooth’W(2) ‘exactly’W(3) scalar particle intensifying – selfW(2) ‘even’ ‘through to the end’W(2) ‘even’
When looking at these semantic stages, one can not but wonder what kind of absentmindedness or creative impulse would drive a speaker – any speaker – to initiate the respective innovation. Scalar particles notoriously have incensed researchers’ interest (for a survey, see, e.g., Traugott, 2006). Consider fast2 to fast3 in German. Who would use a word meaning ‘very much’ to express the concept of approximation? Take even: Who would use a word that means ‘exactly’ to express the concept of scalar extremity? In the examples below, I try to give a feeling for the distance between old meaning – new meaning. If we were to witness a change scenario, speaker of Lold would have utter the (a) sentence and the hearer would have to understand something like the (b) sentence. (2) a. b. (3) a. b.
speaker: ‘‘Tom is very drunk’’ hearer understands: ‘Tom is almost drunk’ speaker: ‘‘Sally went exactly to the police’’ hearer understands: ‘Sally went even to the police’
24
Regine Eckardt
It is virtually impossible to conceive of any context/content of utterance where the (a) utterances would give rise to implicatures like those in (b). This is particularly clear for the case of fastGerman. It has been argued that almost S entails not S (Sadock, 1981) while very S certainly does entail S. The relevant entailments are summarized in (4). (4) Tom is very drunk-Tom is drunk. Tom is almost drunk-Tom is not drunk. If (2.b) were an implicature of (2.a), we’d face a case of an utterance with a logically contradicting implicature. This only happens when a speaker flouts the maxim of quality and makes an ironic statement. It would be possible, of course, to use very/fastGerman in an ironic statement. Yet, then we’d normally understand that ‘‘very much P’’, ironically, conveys that ‘‘not P at all’’. For the other three instances of change, it is likewise hard to tell a story how implicatures should give rise to the newer sense. Yet, speculations are of limited value and instead of debating the possibility or impossibility of certain types of implicatures, we should take a look at usages of the items in question that have been found indicative for imminent or ongoing change by earlier authors. Traugott (2001, 2006) offers a detailed discussion of data in the phase of emergence of even in the scalar sense. Among other examples, she offers (5) as an interesting quote around the turning point from ‘exactly’ to scalar particle. (5)
when I remembre your ffavour and your sadde loffynge delynge to me wardes, ffor south ye make me evene veray glade and joyus in my hart; [ . . . ] ‘when I remember your beauty and sober loving behaviour toward me, truly you make me oevenW very glad and joyous in my heart . . . ’ (1476 Private Letters of John Shillingford, II, 7 after Traugott, 2001: 10)
We will certainly agree with Traugott’s diagnosis that (5) does not show a straightforward use of even in the then predominant sense ‘exactly, just’. However, it is not a use of the type that one would expect in the light of the GIIN theory either. Specifically, this is not a passage where the speaker literally utters ‘‘you make me exactly happy’’ and thereby implicates ‘‘you make me very happy’’. As the first proposition does not implicate the second even in particular, it can’t be a generalized invited implicature, either. Pre-theoretically speaking, (5) simply looks like a mistaken choice of words by the writer, a Thomas Betson to his cousin Katherine Ryche (Traugott, 2001: 10). Traugott glosses the use as ‘‘emphatic’’ which is a plausible prose characterization of the passage, but not part of an analysis of the development in terms of GIINs. Similar ‘‘mistaken’’ uses can be found for the other three items in (1) at the turn between older and newly emerging additional sense. Given that they don’t seem to exemplify implicature, nor irony, nor any other known rhetorical pattern, one might want to know what was going on there. I will elaborate the hypothesis that such examples show instances of pragmatic overload and are
Avoid pragmatic overload
25
reinterpreted by the reader in order to Avoid Pragmatic Overload (APO). In the next section, I will introduce the notion of presuppositions in some more detail before we turn to an illustration of the APO reanalysis on basis of our four sample items.
3. PRESUPPOSITION
IN
HISTORICAL LINGUISTICS
The presuppositions of a sentence S are those pieces of information that the speaker needs to believe in order to make sense of S: S presupposes f if S can only be reasonably be uttered if f is assumed to hold true. Notably, presuppositions need to hold independently of whether the content of S is asserted, denied, questioned, modalized, etc. (see Geurts, 1999 for a very clear survey of presupposition tests). A presupposition f of S is not entailed by S; negating S, or asking whether S, also requires that the speaker believes that f. The following sentences illustrate the phenomenon. (6) My grandmother stopped smoking pot. (7) My grandmother did not stop smoking pot. (8) Has your grandmother stopped smoking pot? Each of (6)–(8) presuppose f ¼ ‘My grandmother used to smoke pot.’ Ideally, in actual communication, speaker and hearer share information that is presupposed by the speaker’s utterance. For the speaker, the requirement is tantamount to not making senseless contributions. A rational speaker will only assert S if s/he believes that the presuppositions of S hold true. An utterance like the following is incoherent. (9)
x My grandmother stopped smoking pot and I don’t believe that she ever smoked pot.
For the hearer, matters may be somewhat different. In many cases, the hearer will not have been aware of all pieces of information that are presupposed by the speaker’s utterance. Having spotted the presuppositions, however, the hearer will usually assume that the speaker has made a meaningful contribution and adopt the presuppositions as part of shared common knowledge. In technical terms, the hearer will accommodate the presuppositions of the speaker’s utterance. (10)
A: Did Granny finally stop smoking pot? B accommodates: ‘A’s grandmother must have been smoking pot.’
Stalnaker (2002) offers an explicit modeling of presupposition accommodation in terms of common ground update. Presupposition accommodation is exploited rhetorically, for instance when the speaker wants to convey information without plainly asserting it. When a teacher tells her student ‘‘I regret to inform you that you have failed the exam’’, she actually asserts ‘I regret S’ and presupposes ‘S holds true’. The student will have to accommodate S: ‘I failed the exam’ in
26
Regine Eckardt
order to make sense of the teacher’s assertion ‘I regret S’. Hence, the teacher will effectively, but not rhetorically, have informed the student about the failing. Yet, such rhetorical tricks will not play a role in our examples. Sometimes it may be hard to guess and accommodate the correct presuppositions that the speaker has in mind. In such cases, the hearer can ask back. In (11), the particle also gives rise to the presupposition that Tommy knows more persons who wear wonder bras. If Sue does not know who that may be (Tommy himself being an unlikely option), she can ask back. (11)
Tommy: Do you also wear ‘wonder bra’? Sue: Yes, why – who else does?
Finally, the presuppositions of the speaker’s utterance may be problematic in that they are in conflict with general knowledge. Assume that someone utters (12) (with the indicated accent, and the additive particle associating with you). (12)
Are you
ALSO
a mother of Peter Smith?
In (12), also gives rise to the presupposition f ¼ ‘the speaker knows more mothers of Peter Smith’. Most likely, the hearer would challenge the presupposition (‘‘Peter Smith has several mothers?’’) and reject it. Sometimes, as we will see, such unreasonable presuppositions are less easy to express than in the case at hand. In such cases, hearers are less likely to start debating and instead may just try to make sense of the utterance one way or other. Even in (12), the hearer could decide to adopt an interpretation of the sentence that allows for the accommodation of the presupposition. Specifically, the word mother could be interpreted in a way that allows for people to have several mothers (‘‘interpret mother as one of the women who pamper Peter Smith’’). However, the more farfetched such interpretations get, the more semantic charity is necessary in order to do justice to the presuppositions. This is what I call ‘‘the utterance carries a pragmatic overload’’.
4. IMPLAUSIBLE PRESUPPOSITIONS We will now turn to our list of four items even, fast, perfino, and selbst. It turns out that the items in their older senses likewise give rise to presuppositions, but to more subtle ones than the examples in Section 3. We will take a closer look at each case in turn. At the time antedating the emergence of the contemporary senses of these words, we find uses of the word in sentences where presuppositions can not be accommodated because they contradict common world knowledge. Let us say that little pragmatic ‘‘accidents’’ happened from time to time. Not all of these accidents necessarily need to lead straight to the newly emerging sense of the word. However, many of these pragmatic ‘‘accidents’’ could best be repaired by the hearer if he or she hypothesized that the problematic item was in fact used in a new sense. In such cases, the modern
Avoid pragmatic overload
27
reader will be able to interpret the sentence on the basis of the modern use of the word, and the only remaining surprise would be how the contemporary reader/hearer would have been able to do the same, given that he did not know as yet where language history would lead. The present section illustrates pragmatic accidents for fast (German), even, perfino, and selbst. In each case, I will briefly specify the older reading of the word, spell out the presuppositions of that older reading, and then show examples from the crucial time of change where these presuppositions were hard or impossible to accommodate.
4.1. fast (German) German and English fast go back to a common adverbial root which meant ‘‘tightly, fast’’ in the sense of both physical attachment as well as mental attachment to a cause. It is used in this sense in the examples in (13) and (14) (all quotes from Deutsches Wo¨rterbuch, Grimm and Grimm (1854–1960)). (13)
so¨lh pflicht halt fast ‘this duty hold fast’ (1535 Schwarzenberg 139, 2, from DW 3:1348)
(14)
halt fast den pfluog ‘hold the plough fast/tightly’ (1535 Schwarzenberg 140, 2, from DW 3: 1348)
Subsequently, the notion of taking a ‘good grip’ was extended to gradable properties in general. The word became a degree modifier with the meaning of very much. Here, German and English part ways, the English adverb adopting the notion of moving on with ‘high speed’, as laid out in the classical study by Stern (1931). The German degree use is illustrated in (15) and (16). (15)
dis ler und trost mich fast erquickt ‘this lesson and consolation revives me very much’ (1535 Schwarzenberg 152, 2, from DW 3: 1348)
(16)
wenn du gleich fast darnach ringest, so erlangestu es doch nicht. ‘even if you struggle for it hard, you will not attain it’ (1534 Luther, Sir. 11, 2, from DW 3: 1350)
The use of an adverb which reports that a property P held to a high degree carries the presupposition that P is a gradable property in the first place. Otherwise, the combination fast P will not make any sense. The DW offers quotations like the following, around the time when the approximative use emerged. The authors of the dictionary clear-sightedly comment that here, the sense of fast was
28
Regine Eckardt
‘‘leaning towards the newer sense ‘almost’’’. Taking a closer look at the respective examples in order to understand why this may be so, we note that they typically fail to combine fast with a gradable property. (17)
weil er fast hundert ierig war ‘he was very much?/almost? hundred years old’ (1534 Luther, Ro¨m. 4, 19, from DW 3; 1350)
(18)
kamen darauff fast um zwo uren ‘(they) arrived there very much?/almost? at two o’clock/sharp?’ (c. 1576 Fischart gl. schif 185, from DW 3: 1350)
(19)
Nun war gedachtes VerzeichniX so accurat eingerichtet, daX fast nicht ein Balcken vergessen war, ‘that very much?/almost? not ONE log forgotten was’ wo er solte eingeschoben, wie er solte bekleidet oder gemahlet, wie er solte behobelt und beschnitzet werden. (1672, Weise erzn. eingang, from DW 3: 1350)
We will discuss this impression of ‘‘leaning towards a newer sense’’ on the basis of an utterance like (17), rendered in the English equivalent in (17u). (17u) He was very muchdeg 100 years old. Be 100 years old is not usually something that one can be with more or less intensity. Either the speaker was unaware of the problem and erroneously chose fast instead of the qualifier that expresses what he actually had had in mind. Or the speaker indeed conceptualized the property be 100 years old as gradable; perhaps thinking of degrees of senility, or wisdom, or poise. The reader will notice the problem: very much presupposes a gradable property, but be 100 years old isn’t one. The required mental search for a way in which be 100 years old could possibly be conceived of as gradable is what makes (17) hard to process: it creates the pragmatic overload for the utterance in (17). For completeness’ sake, note that (17) can not give rise to any implicatures or other inference (general or particular) unless the reader finds some way to map the utterance onto some literal content in the first place, because meaningless utterances don’t give rise to implicatures. If this first derivation of literal content should already carry the reader to the proposition ‘he was almost 100 years old’ then the change from very much to almost does not arise by implicature. In principle, of course, it could be claimed that the reader computed some other proposition q as the literal content of (17) and that this other proposition q, in that context, implicated the newer sense ‘he was almost 100’. But then, I can’t see what q could possibly be.
Avoid pragmatic overload
29
4.2. even In its earliest attested stage, once again shared by the German cognate eben, the word even denoted ‘evenly, smoothly’ as a property of surfaces. (20)
Do past or cleye ther-upon al aboute as ytold bi-fore, caste Scalding hot honey euene ther-upon ‘Put paste or mud thereon all around as said before cast scalding hot honey evenly thereon’ (c. 1450 Horses, p. 113 [Helsinki], quote/translation after Traugott, 2006: 346)
This notion can also be applied in cases where two or more objects put together form an even surface, and hence one fits the other evenly. From such uses, a somewhat later sense developed that can be paraphrased in modern English by: ‘exactly, precisely; in (exactly) equal degree’. As to be expected, the exact match can be one between two objects, or an object and a measure, as illustrated in (21) (the passage describes the measures of Noah’s ark). (21)
The heght is euen thyrty Cubettys full strenght. ‘the height is exactly thirty cubits full strength’ (c. 1500 Towneley Plays, p. 21 [Helsinki], quote/translation after Traugott, 2006: 347)
The qualification that something was ‘‘exactly, just, precisely P’’ presupposes
a topology of approximating P-hood (canonically illustrated by numbers and scales like ‘roughly 20 years old’ – ‘exactly 20 years old’)
approximation from more than one direction
The topology need not rest on scientific scales or geometrical spaces; human concepts can rest on more general topological spaces (e.g., Ga¨rdenfors, 2000). The examples in (22) rest on a topology of similarities between singing events, and report closer and loser similarities between events where nightingales are singing and where humans are singing. (22)
She sang approximately like a nightingale She sang exactly like a nightingale
The notion of ‘exactness’ is inapplicable
when a property P can not be approximated (#be roughly/exactly pregnant)
when a property P is inherently vague itself (# be roughly/exactly angry)
when a property P is the polar end of a scale
The last restriction may come somewhat surprisingly, but can easily be verified if we consider infelicitous examples like those in (23). The infelicity arises for all adverbials that express
30
Regine Eckardt
‘‘exactness’’, and can be reproduced in other languages as well. Hence I take it that the restriction is not one that only accidentally applies to one special adverb of contemporary English alone. It appears to be part of the notion of an exact hit, and its opposite, the missing, that the target point can be missed in more than one direction. (23)
a. b. c.
#Tom exactly emptied the glass. #The dog exactly died. #Anna read the book exactly to the end.
This was the state of English even at the time when the scalar particle started to develop. The following quotes, taken from Traugott (2001) alone, show attested uses of even around the time of the emergence of the scalar particle. Each of them fails to satisfy one or the other aspect of the topology that is presupposed by the notion of an exact (‘even’) match. (24)
whanne I remembre your ffavour and your sadde loffynge delynge to me wardes, ffor south ye make me evene veray glade and joyus in my hart: and on the tothersyde agayn whanne I remembre your yonge youthe. And seeth well that ye be none eater of your mete, the which shuld helpe you greately in waxynge; ffor south than ye make me very hevy again. ‘When I remember your beauty and sober loving behaviour toward me, truly you make me really very glad and joyous in my heart, and on the other hand again, when I remember your young age, and see clearly that you are no eater of your food, which should help you greatly in your growing, truly then you make me very sad again.’ (1476 Private Letters of John Shillingford, II, 7, in Kingsford’s Stonor Letters and Papers; translation after Traugott, 2001: 10)
What we see here is the combination of evene and veray glade (‘truly happy’). The latter denotes a gradable property that has vague boundaries. There are no clear criteria from where on a person should count as veray glade in contrast to ‘‘simply’’ glade. Against this background, evene is supposed to contribute some qualification of J.S.’s happiness. The reader will face a conceptual mismatch between the modifier evene and its argument veray glade, which violates the topological properties presupposed by evene. The passage is pragmatically problematic if we assume that the author attempted to use a word that means ‘precisely’. We could conceptualize the property veray glade as one with clearly circumscribed extension, but it is not clear what should define the exact boundaries of happiness. We therefore face a pragmatic overload. Traugott boldly glosses even as ‘really’, granting J.S. reasonable (though innovative) use of (Early Modern) English and thus nicely illustrates the move of the charitable hearer. The passage is classed as ‘‘must be emphatic’’, and indeed, emotional undertones seem to carry away the writer. It is unclear, however, whether
Avoid pragmatic overload
31
we don’t just witness a use of the older item even ‘exactly’ with poorly justified presuppositions.1 The following quote of Traugott may show another poorly justified use of ‘exactly’. (25)
is not this he that sate and begged? Some sayde: this is he. Other sayd: he is lyke him. But he him selfe sayde: I am even he. ‘Is not this the man that sat and begged? Some said: This is he. Others said: He is like him. But he himself said, I am exactly (?) he.’ (1534, Tyndale, New Testament, IX, i. quoted and translated after Traugott, 2006: 349)
The protagonists are concerned with the similarity of one person to another. Taking into consideration what the dialogue is actually about, the notion of exactness seems once again poorly in place. To be or not to be identical to someone is, after all, a categorical property and not one that can be approximated: ?Saulus is roughly/exactly identical to Paulus. What might have interfered is the property of looking similar/exactly like someone. Of course, we can think of visual similarity to ever higher degrees. But that is not what is at stake, the interlocutors do not care whether the speaker is a twin of the man that sate and begged. The passage in (26) shows another use of even where the presuppositions of the word in its older sense are violated. (26)
but sayde, he had rather be sycke even vnto death then he wold breake his espousals. ‘But said he would rather be sick precisely?/even?/ø unto death than break his vows.’ (1449 Latimer, Sermons: 36 after Traugott, 2001: 11)
Once more, the use of even fails to adhere to the presuppositions of approximation. While ‘dying’ can be approximated by sicknesses of different degrees of severity, it is conceptually impossible to ‘‘miss’’ dead and die ‘‘roughly’’ in contrast to ‘‘exactly’’. The reader at the time faced the task to accommodate an impossible presupposition, which reads roughly like: ‘‘death is not the polar end of the scale of sicknesses of increasing severity’’ (‘exactness’ presupposing that the exact hit can be missed in several directions). This quotation will quite naturally lend itself to a reanalysis in the scalar sense that is expressed by modern even. Note that the range of attested uses of even illustrates that not all the pragmatic ‘‘accidents’’ need to receive a rescue interpretation that leads toward the modern sense of the word. In retrospect, all we can say is that those rescue interpretations, which led to the hypothesis that even denoted the scalar particle determined the future use of the word. Of course, it can not be proved that the specific passage in (26) (or any other, at that) is the driving pragmatic accident. However, it can plausibly be assumed that the crucial passages looked somewhat like (26).
1
It is certainly right to assume, as Traugott does, that J.S. may have intended to say ‘‘really very happy‘‘, but the words chosen may not have expressed this thought even at the time.
32
Regine Eckardt
Let me finally point out, again, that neither (24) nor (25) support the assumption that even in the scalar sense emerged by generalized implicatures (GIIN analysis). In neither (24) nor (25) is even used in its literal old sense plus giving rise to an implicature that exactly P was most surprisingly P. The word even is not used reasonably in its old sense at all, and hence can not give rise to implicatures. The reader’s hermeneutic activity is initiated by the very observation that the older use of even can not be the one at stake in either example. 4.3. perfino In its modern sense, perfino translates to English even and can be used in sentences like the one in (27). (27)
E` venuto perfino Matteo ‘Even Matteo came’ (Visconti, 2005: 238)
In its traditional sense in Old Italian, perfino literally meant ‘through until, to the end’. A typical example, offered in Visconti (2005), is shown in (28). (28)
b. Dentro in un bosco, che’e` quivi vicino,/t’ imbosca es sta perfino al mattutino. ‘In a wood, which is near here, hide yourself in the wood and stay until morning’ (–1380 La Spagna after Visconti, 2005: 243)
A word that reports that P ‘‘holds through until the end’’ presupposes a topology, which defines betweenness and the ancillary notion of an uninterrupted state between two points. The topology can be one in space, but also one on more abstract domains. The reached point can be neutral (like in (28)) but also conceptualized as a polar endpoint on a scale. (29)
In cio` ancora che persevero` in croce perfino alla morte, ci da` ammaestramento di perfetta obbedienza e pazienza, e di perseverare nella penitenza ‘In that too, that he endured the cross until death, he provides us with an example of perfect obedience and patience.’ (1432 Cavalca, Esp. simbolo after Visconti, 2005: 243)
The following quote around the time of emergence of the scalar use ‘even’ is taken from Visconti (2005). It fails to warrant the presuppositions of perfino in its older sense. (30)
[ . . . ] in acqua, in neve, in grandine o pruina: a tutto il ciel s’inclina, perfino a quel che la natura sprezza’ ‘Water, snow, hail or frost: To everything bends the sky, even to that which nature despises’ (1389–1420 S. Serdini, Rime after Visconti, 2005: 244)
Avoid pragmatic overload
33
The use of perfino in (30) presupposes the existence of a continuum, which passes from water, snow, hail, and frost to ‘‘that which nature despises’’. Or, is frost the thing that nature despises? And, is there a continuum from good to ever more evil things, or isn’t it more a categorical distinction between ‘‘the good’’ and ‘‘the bad’’? Such questions are indicative for an utterance with a presupposition, which may or may not be plausible. In some sense, the speaker in (30) is free to believe in continuous scales of whatever kind (mathematically, there is practically nothing that could stop you from ordering any set whatsoever). But if the scale is too ad hoc, or too private, the hearer will have a hard time in understanding how the speaker believes the presuppositions of (30) satisfied. S/he has hence the choice between guessing and accommodating the speaker’s scale ( ¼ what the speaker believes hearer will do) and accepting a presupposition failure due to lack of a suitable scale. If the hearer decides for the latter option, s/he will next hypothesize that the speaker actually used perfino in a different sense, and interpret the utterance (30) in a new way – possibly and plausibly assuming that perfino was meant to mean ‘even’. 4.4. selbst Like its English cognate, selbst is attested at an early stage as an intensifier in German. It can still be used in this sense, in the form selbst and its variant selber, which unambiguously denotes the intensifier. (31) shows an example. (31) Gott selber ruht sich manchmal aus. ‘God himself takes a rest sometimes.’ The intensifier associates with a certain nominal element in the sentence. (31) shows an adnominal use where the intensifier follows the associated element directly. As argued in Baker 1995, Kemmer 1995, the use of selber/selbst presupposes that the associated object (here: ‘God’) is conceptualized as the center of a salient entourage (here: ‘God’ as the center of His creation). Generally, the use of intensifying selber only makes sense if the associated referent, the x of x-self so to speak, is understood to come with an entourage, peripheral objects or persons, in which x constitutes a central point. I will refer to this presupposition as the entourage presupposition; we will leave the details on focusing and accent placement aside for the ease of exposition. (32) shows a quotation from 1650, around the time when the scalar use emerged. We can assume that the newer sense selbst ¼ even was not in use at the time when this passage was written. I offer the full context, but the crucial part is the underlined sentence at the end. (32)
Man kan/es ist nicht ohn/ein blut begierig Thier Gewo¨hnen daX es spiel vnd nieder knie vor dir/ Man kan/waX noch viel mehr/die starcke flut vmbkehren. Den stro¨men widerstehn/den tollen wellen wehren.
34
Regine Eckardt
Man da¨mpfft der flammen macht/man segelt gegen wind/ Man stu¨rtz’t die felsen hin wo tha¨l vnd ho¨len sind. ‘One can, it’s not easy, a bloodthirsty animal/train so that it will play and kneel down before you/One can, which is even more, reverse the strong flood/resist the streams, restrain the wild waves/One damps the mighty flames, one sails against the winds/One throws boulders where there are valleys and caverns.’ Man kan die steine selbst mit weitzenu¨berziehen. ‘One can the stones ‘selbst’ with wheat cover’ ‘One can cover the stones ‘selbst’ (themselves/even?) with wheat.’ (1650 Leo Arm., II, 5) In a traditional interpretation of the passage on basis of the intensifier, the reader ends with the following ingredients for the eventual meaning of the underlined sentence: (33)
selbst ¼ intensifier, is supposed to associate with Steine ‘stones’ to yield ‘stones themselves’ presupposition: ‘stones’ are something that has an entourage, that is the natural center of some ontological domain.
Once again, this presupposition is not logically contradictory but still hard to satisfy or fill with content. What could be the entourage of stones? Mentioned in the text are blutbegierig Thier ‘bloodthirsty animal’, starcke flut ‘flood’, tolle wellen ‘wild waves’, flammen ‘flames’ – but ‘stones’ are not a plausible center in this entourage of things either today or in 1650. To check this, consider the following statement: (34)
Bloodthirsty animals, high flood, wild waves, flames: yet, stones are the worst of them all.
This assertion is as implausible today as it must have been implausible in 1650. Once again, the reader will face a pragmatic overload when s/he tries to work out an appropriate way to accommodate the presuppositions of (32) under the old reading of selbst. Once again, this may be reason for the reader to wonder whether the speaker intended to use selbst in a new sense, one that would not give rise to unaccommodatable presuppositions. We have now seen several instances utterances with pragmatic overload, caused by items that soon after adopted a new meaning. In the next section, I will discuss in more detail why the repair strategies for the quoted examples could indeed lead the reader/hearer to assume that the word was used in a new sense, and that it was plausibly the sense that we find in permanent use some decades later. Before moving on, let me mention that another nice instance of presupposition failures in the pre-change phase can be traced in the development of lauter, see Eckardt (2006: Chapter 7). What is most striking about such observations, however, is that pragmatic accidents of this kind clearly are not restricted to language use in ancient times. Instances of pragmatic overload can be observed on a daily basis as soon as you start watching the utterances in your daily
Avoid pragmatic overload
35
environment. It is tempting to think that repair interpretations can in some cases converge on recognizable new stages in the use of words.
5. APO,
AND THE
CHARITABLE HEARER
The examples in Section 4 illustrate how an utterance can create a pragmatic overload: The utterance under a conservative interpretation will trigger presuppositions that the hearer can not easily accommodate or refute. The hearer has three options: S/he can (a) be uncharitable and refuse to interpret the utterance at all, or (b) face the pragmatic overload and attempt to reconceptualize the world such that the presupposition makes sense and is consistent, or (c) hypothesize a new meaning for parts of the utterance, notably the item that gave rise to the problematic presupposition. Option (c) will allow the hearer to Avoid the Pragmatic Overload. In the examples at hand, it makes indeed sense to believe that (c) is a viable option for the reader. The hypothesized interpretations for fast, even, perfino, and selbst which a posteriori turned out to be adopted as new meanings into the lexicon will fit the context better than the conservative senses, and deviate from the older senses in only minimal ways. We will take a look at each case in turn. fast: Combining fastold with a non-gradable property P creates the problem that the degree modifier can not contribute semantically. The repair strategy was to look for a superproperty P of P which in turn is gradable, and where P is the polar end point. (In the case of being 100 years old, the scale of lower ages offers itself, for example.) The semantic contribution of fast was then taken to be relative to this gradable superproperty: The subject has a property which ranks high on the P scale. (e.g., the subject has an age that ranks high on the 0–100 year scale; in other words, he is almost 100 years old.) (35) very much 100 years oldWalmost 100 years old
property P (gradable)
very-P
Figure 2.1. Fast ¼ very.
(gradable) superproperty ∏
Figure 2.2. Fast ¼ almost.
almost-P
P (nongradable)
36
Regine Eckardt
Interestingly, the semantic contribution of fastnew remains the same as for fastold, but the presupposed background has changed drastically. Figures 2.1 and 2.2 visualize the change. The change implements the idea that (36)
very P ¼ almost P
Eckardt (2007) and similarly Penka (2006), offer a full semantic spell out of this idea. The proposed semantic analysis for fastnew not only is plausible in terms of language history, but moreover covers all possible uses that we witness in modern German and constitutes an improvement on earlier analyses that essentially rest on Sadock (1981). In view of the appealing simplicity of the reanalysis in question, it becomes all the more implausible to believe that the semantic change occurred on the basis of some hitherto undetected kind of use where fast/very much could give rise to implicatures in the sense of fast/almost. even: In the case of even, the adjustment in the mental lexicon was again minimal. As discussed above, even in its older sense ‘precisely’ will have required – like all other adverbs that express the notion of an exact hit or match – that the state of affair Q in question is part of a topology of more or less similar states of affairs surrounding it. Importantly, the notion of ‘exactness’ prohibits topologies where Q is the polar end point of a linear topological space.2 This prohibition was violated in examples like (26), repeated in a modern gloss in (38). (37)
exactly fatally sick
The hearer accepted that evenold was used in a situation where it applied to a polar end point. It would, however, be void to adopt a word that lexifies what is obvious from the content of the rest of the sentence. The underlined part would explicate the semantic contribution: The subject could be sick to death, and this is as far as you can get. Hence, the hearer hypothesized that the scales in question correlated with a more subjective scale of the speaker, for example, one of increased surprise, decreased probability, (here) increased woe, etc. These subjective scales formed the backbone of modern even in the sense exemplified in (39). (38)
sick even to death
Two points may be worth mentioning. First, these interpretive efforts on the hearer’s side might often be at the basis of subjectification in the sense of Traugott (1995). The hearer, confronted with (38), has to make a guess what the speaker might have wanted to express. In many cases, there are canonical answers to this question. At least in the cases at hand, we do not face a
2
A topology, in mathematical terms, is a space with a measure function that measures the distance between any two points in the space. See Ga¨rdenfors (2000) for a general theory of concepts in topological spaces.
Avoid pragmatic overload
37
speaker-driven process. Uttering incoherent sentences and hoping that the hearer will grab your message does not seem a rational communicative strategy.3 Second, the shift that leads from (38) to (39) can not be described as implicatures of an interpretation of (38) on the basis of even in its older sense. Nor can I think of any uses of even ‘exactly’ that create an implicature ‘even’, an implicature that could then be reanalyzed as literal meaning (GIIN).4 The pragmatic overload view on change, in contrast, can offer a plausible analysis for attested examples rather than speculating on the existence of unconceivable unattested cases. perfino: The shift of perfino concerns once more the kinds of scales that can be referred to. Traditionally, perfino was restricted to scales in the world, as was shown in Visconti (2005). The pragmatic accident happened in cases where the hearer could not make out any reasonable conceptual scales. In (30) above, the writer lists a number of evils but does not suggest that they are linearly ordered in any way. I repeat the example in the English version, volitionally inserting the translation of the older sense of perfino. This is the semantic material for the (earlier Italian) reader to start with. (39)
the sky bends ‘‘through until’’ to that which nature despises
The underlined part presupposes a scale. Lacking a more contentful continuous scale, the reader can meet the need for a scale by interpreting the sentence against the all-purpose scale of increased surprise/decreasing probability. The reader will hence understand that the proposition ‘the sky bends to that which nature despises’ is presented as the endpoint on that scale. Particles that signal this kind of side information are those that correspond to ‘even’. Hence, (40) with the surprise scale is synonymous to (41). (40)
even to that which nature despises
selbst: In the turning point examples of selbst, the hearer/reader was faced with the task to reconstruct a core–periphery structure that is presupposed by intensifiers (-self). As shown in the above quotation, it may not have been inconsistent, but simply implausible to adopt such a core– periphery for the examples at hand. (42) would require us to think about stones as the core instance of nonfertile ground (true) after a long passage that did not have to do with fertility in the least, but rather with various kinds of attempts to fight hostile natural forces. (41)
3
grow wheat on [stones themselves]
Notwithstanding that of course, incoherent linguistic behaviour and emotional stress often go hand in hand. But incoherence in such situations is rather taken as a metalinguistic signal by the observer, usually. 4 This does not, of course, mean that such uses do not exist. Maybe it’s just that I am not trying hard enough. Yet, I have not so far encountered any plausible fictitious example where any sentence with a word that means ‘exactly’ would implicate the analogous proposition with ‘even’ replacing ‘exactly’.
38
Regine Eckardt
In such situations, the reader may have refused to face the pragmatic overload and instead searched for a simpler interpretation of the sentence. While it is conceptually hard to switch from tempests, floods, and flames to soils of different quality, it is quite easy in the given context to understand the tasks as ordered to be more and more unfeasible. The task that can least likely be performed is the task marked by selbst. This hypothesized semantic analysis will have several consequences. The new item selbstnew can take sentence scope, it no longer is in focus (unlike in the older construction; see Eckardt, 2001), focusing may be used to indicate the kind of alternatives on the scale, and selbstnew indicates that the reported state of affairs is high on that scale. The resulting reanalysis is the one in (43). (42)
even [grow wheat on stones]F
In the present overview, I omit the details of the meaning shifts at stake (see earlier work for detailed semantic analyses). Case studies reveal that hearers are surprisingly conservative, maintaining as much of earlier semantic structure as possible and making new content fit. The present examples suggest that the concept of scale is important in human thinking both as a new player in meaning (even, selbst, perfino) as well as a semantic part to be maintained ( fast). Let me sum up some characteristics of language changes that arise in the attempt to APO.5 First, they constitute instantaneous changes of language in the hearer’s competence, resulting from the attempt to interpret one specific utterance. Whatever gradual spreads may occur afterwards – the hearer gradually gaining faith in a new sense of a word, his surroundings being gradually infected to share this new item – the initial shift is discrete. The competence of the speaker may be momentarily weak, but all in all we face competent speakers without attempts at creativity. The witnessed lapses of tongue or pen are not beyond what we see daily in our own environment, interacting with speakers that we hold fully competent. The competence of the hearer likewise need not be doubted. (In some sense, s/he is still the more competent of the two.) The reanalyzes do not fit under the label of ‘‘misunderstandings’’ of the kind prevalent in other domains of language change, for instance contact phenomena or L2 speaker errors. From a more general perspective, the trend to APO bears resemblance to similar trends toward simplicity and clarity that have been postulated in syntax, specifically in language acquisition and
5 I am not claiming that all language change is instantaneous, that all language change is initiated by the APO principle, that all language changes of a certain class – e.g., grammaticalization – are initiated by APO. The more modest claim is simply that instantaneous change does exist. It does not rest on stupid or creative speakers. It does not, in all likelihood, leave an impression of novelty in the hearer. It is part of perfectly normal ordinary use of language (Hermann Paul).
Avoid pragmatic overload
39
language change. Lightfoot (1979) is a classical reference. Lightfoot attempted to explain the development of the modal system in English by, among other factors, the dismissal of more complex structures in favour of simpler options. Lightfoot’s ‘‘avoid complex syntactic structure’’ principle in 1979 can be seen as the syntactic twin principle to ‘‘APO’’ on the semantic/pragmatic side. Even though Lightfoot reshaped his ideas about the development of the modals in later years, the idea that simpler syntactic structures replace more complex structures in developmental processes has been retained (e.g., Lightfoot, 1991, 1999). The APO analysis is moreover appealing in that it fully adheres to the uniformitarian principle. The uniformitarian principle in language history states, roughly, that any analysis of language change should rest on processes that can still be observed in contemporary language use. The APO analysis fulfils this requirement. The proposed changes rest on communicative situations that we frequently witness ourselves. What is still unexplored is a quantitative analysis of this type of change. How many pragmatic accidents are necessary for change to become likely? What is the ratio of accidents in total to accidents where reanalysis led to the item that was later adopted into the lexicon? So far, these questions can only receive an impressionistic answer, as traditional classifications tend to sort quotes of words by sense, not by more general categories like whether they create pragmatic overload. Many studies report that the point where a new item emerges is frequently preceded by a phase of great variability in the use of the crucial word. We might understand these pre-phases as times where a word was promoted by sociolinguistic factors (‘‘fashion words’’) or where other priming effects led to more retrievals of the word than what would have been warranted. The modest aim of this paper has been, to offer a new way in which we might understand what really happens, linguistically, in some of the more puzzling ones of such variable uses at the turning point between an older and an emerging new word sense. The appeal of this analysis consists in that it addresses the attested quotes themselves. The attested quotes are perceived as possible source of change. The quotes are not classed as clear instances of the new language use, ‘‘made possible by changes that we see nowhere in the data’’. In that sense, the present approach adheres to my slogan in Section 1: it is one more attempt to understand language change without mysteries.
REFERENCES Baker, C. L. (1995). ‘‘Contrast, discourse prominence, and intensification, with special reference to locally free reflexives in British English’’. Language 71: 63–102. Grimm, J. and W. Grimm, (eds.) (1854–1960). Deutsches Wo¨rterbuch, S. Hirzel, Leipzig, Quellenverzeichnis 1971. Eckardt, R. (2001). ‘‘Reanalysing ‘selbst’’’. Natural Language Semantics 9: 371–412.
40
Regine Eckardt
Eckardt, R. (2006). Meaning change in grammaticalization. An enquiry into semantic reanalysis. Oxford: Oxford University Press. Eckardt, R. (2007). Almost – A theory, Ms, University of Go¨ttingen and Semantics Archive. Ga¨rdenfors, P. (2000). Conceptual spaces: The geometry of thought. Cambridge, MA: MIT Press. Geurts, B. (1999). Presuppositions and pronouns. Oxford: Elsevier. Kemmer, S. (1995). ‘‘Emphatic and reflexive – self: Expectations, viewpoint, and subjectivity’’, in D. Stein and S. Wright (eds.), Subjectivity and subjectification. Cambridge: Cambridge University Press, 55–82. Ko¨nig, E. and E. C. Traugott (1988). ‘‘Pragmatic strengthening and semantic change: The conventionalizing of conversational implicature’’, in W. Hu¨llen and R. Schulze (eds.), Understanding the Lexicon: Meaning. Sense and world knowledge in lexical semantics. Tu¨bingen: Niemeyer, 110–124. Lightfoot, D. (1979). Principles of diachronic syntax. Cambridge: Cambridge University Press. Lightfoot, D. (1991). How to set parameters: Arguments from language change. Cambridge, MA: MIT Press. Lightfoot, D. (1999). The development of language: Acquisition, change and evolution. Oxford: Blackwell. Paul, H. (1920). Prinzipien der Sprachgeschichte. 5th edn. Halle: Niemeyer. Penka, D. (2006). ‘‘Almost there: The meaning of almost’’, in C. Ebert and C. Endriss (eds.), Proceedings of Sinn und Bedeutung 10. ZAS Papers in Linguistics 44, 275–286. Sadock, J. M. (1981). ‘‘Almost’’, in P. Cole (ed.), Radical pragmatics. New York: Academic Press, 257–271. Stalnaker, R. (2002). ‘‘Common ground’’. Linguistics and Philosophy 25: 701–721. Stern, G. (1968 [1931]). Meaning and change of meaning. Indiana University Press, IN. Traugott, E. C. (1988), ‘‘Pragmatic strengthening and grammaticalization’’, in S. Axmaker, A. Jaisser and H. Singmaster (eds.), Proceedings of the 14th Annual Meeting of the Berkeley Linguistic Society, Berkeley Linguistic Society, Berkeley, 406–416. Traugott, E. C. (1995). ‘‘Subjectification in grammaticalization’’, in D. Stein and S. Wright (eds.), Subjectivity and subjectification. Cambridge: Cambridge University Press, 31–54. Traugott, E. C. (2001). ‘‘How do scalar meanings arise?’’ LSA Institute Collitz Lecture, July 18th, 2001. Traugott, E. C. (2006). ‘‘The semantic development of scalar focus particles’’, in A. van Kemenade and B. Los (eds.), The handbook of the history of English. London: Blackwell, 335–359. Traugott, E. C. and R. B. Dasher (2002). Regularity in semantic change. Cambridge: Cambridge University Press. Visconti, J. (2005). ‘‘On the origins of scalar particles in Italian’’. Journal of Historical Pragmatics 6 (2): 237–261. von der Gabelentz, G. (1891). Die Sprachwissenschaft. Ihre Aufgaben, Methoden, und bisherigen Ergebnisse, Weigel, Leipzig.
HISTORICAL
SOURCES
[Fischart gl. schif ] Fischart, Johann (1576). Das glu¨ckhafft schif von Zu¨rich. Nachrduck in Johann Fischart: Geschichtklitterung (Gargantua). Du¨sseldorf 1963. [Leo Arm.] Andreas Gryphius (1560). Armenius: Trauerspiel von Andreas Gryphius. Ed. Peter Rusterholz. Leipzig, Reclam, 1971. [Luther, Ro¨m.] Martin Luther (1534). Biblia. Brief an die Ro¨mer. Complete facsimile edition, Ko¨ln, Taschen, 2002.
Avoid pragmatic overload
41
[Luther, Sir.] Martin Luther (1534). Biblia. Jesus Sirach. Complete facsimile edition, Ko¨ln, Taschen, 2002. [Schwarzenberg] Schwarzenberg, Joh (1535). Von, gebu¨hrliche werk. verdeutschung von Cicero de Officiis. [Weise erzn]: Weise, Christian. 15 Die drey a¨rgsten Ertz-Narren in der gantzen Welt (1672). Nachdruck in Christian Weise: Sa¨mtliche Werke. Berlin und New York 1971.
Current Trends in Diachronic Semantics and Pragmatics Edited by Maj-Britt Mosegaard Hansen and Jacqueline Visconti r 2009 by Emerald Group Publishing Limited. All rights reserved.
3 DIACHRONIC PATHWAYS AND PRAGMATIC STRATEGIES: DIFFERENT TYPES OF PRAGMATIC PARTICLES FROM A DIACHRONIC POINT OF VIEW Ulrich Detges and Richard Waltereit
One of the main concerns of research on pragmatic particles has been the problem of a neat delimitation between the major classes involved (cf. Brinton, 1996; Hansen, 1998a). In this paper, we want to approach this question from a new angle. We will show that the functional difference between two types of pragmatic particles, namely discourse markers and modal particles, can be accounted for in diachronic terms. In particular, we will show that each type is the historical outcome of a different kind of pragmatic strategy in discourse. More broadly, our results imply that there are levels of generalization on semantic change below the overarching tendencies of subjectification (Traugott, 1997; Traugott and Dasher, 2002). At the same time, they provide very specific justification for these levels of generalization, and ultimately for subjectification itself.
1. DISCOURSE MARKERS
AND
MODAL PARTICLES
It is customary to distinguish, in research on pragmatic particles, between discourse markers and modal particles. In what follows, we shall illustrate both functional classes by a look at elements that are derived from adverbs meaning ‘well’ in two Romance languages. Thus, Spanish has a discourse marker bien, which goes back to adverbial bien ‘well’. As a discourse marker, it often occurs turn-initially, as in (1) (cf. Serrano, 1999).
44
Ulrich Detges and Richard Waltereit
(1) Spanish bien as a discourse marker (Corpus Oral, 1992) S1: Mire, yo creo que [ . . . ] los ciudadanos [ . . . ] son muy crı´ticos con el Parlamento porque lo valoran muy alto en lo que es el conjunto del sistema democra´tico del paı´s. La percepcio´n, yo creo que los ciudadanos tienen ah . . . una percepcio´n de que el Parlamento no funciona muy bien [ . . . ], y yo creo que hay que trabajar permanentemente para que los ciudadanos conozcan el trabajo parlamentario [ . . . ]. Pero adema´s de que ese trabajo sea conocido y bien conocido por la sociedad, va en favor, no de los diputados y del Parlamento, sino del conjunto de la convivencia de los ciudadanos que en gran medida depende de que el Parlamento funcione bien. ‘Look, I think that the citizens are very critical of the Parliament because they value it highly within the country’s overall system of democracy. The perception I think that the citizens have a perception that the Parliament does not work very well, and I think we have to work permanently so that citizens know the Parliament’s work. But apart from that, the fact that this work should be known, and known well, by society, benefits not the Parliament and its members, but the entire community of citizens, which largely depends on the Parliament running well.’ S2: BIEN, sen˜or presidente, sen˜or Pons, presidente de las Cortes, nosotros queremos felicitarle desde aquı´ la . . . estas fiestas y las Pascuas [ . . . ]. ‘Well, Mr. President, Mr. Pons, President of the Cortes, we want to congratulate you from here the . . . these festivities and Easter [ . . . ].’ Here, bien does not signal approval of the previous speaker’s argumentation, as it would as an adverbial meaning ‘well’. Rather, it serves to close a topic addressed in the foregoing contribution. At the same time, it can be used to introduce a new topic. This is a function typical of discourse markers: they overtly indicate the relationship of a given chunk of discourse/text to a wider stretch of the same discourse/text (Hansen, 1998a). Put more generally, a typical discourse marker signals a two-place relation which concerns the structure or form of discourse. The scope of discourse markers is extremely variable: what exactly is a ‘‘chunk of discourse’’ is determined on discoursestructural grounds. In particular, the discourse units linked by discourse markers are not necessarily identical with grammatical units at the sentence-constituent level (Hansen, 1998a). In other words, the placement of discourse markers is not subject to constraints of grammatical nature. Unlike Spanish, French has an element bien which diachronically goes back to adverbial bien ‘well’, but which itself is neither an adverb nor a discourse marker. Hansen (1998b) analyzes it as a modal particle. (2) French bien as a modal particle (Hansen, 1998b) Vous avez BIEN rec- u mon message? ‘you have:2PL bien received my message’ ‘You did receive my message, didn’t you?’
Diachronic pathways and pragmatic strategies
45
Table 3.1. Discourse markers versus modal particles Discourse markers
Modal particles
Function at discourse level
Function at speech-act level
Two-place relationship
One-place relationship
Variable scope
Fixed scope
Discourse-structurally governed position
Syntactically governed position
As a modal particle, French bien occurs in assertives as well as in yes/no-questions. In the latter, it indicates that the speaker expects an affirmative answer rather than a negative one. In yes/ no-questions, bien is always restricted to the position following the finite verb. As is illustrated by this example, modal particles function at the speech-act level. Put differently, they are conventionally tied to particular speech-act types. While discourse markers reflect two-place relations within the structure of textual organization, modal particles convey oneplace relations with respect to illocutionary content. Discourse markers have a variable scope, and their position reflects the linear organization of discourse rather than constraints of syntactic nature. Modal particles, on the other hand, are highly content-dependent and have a scope and a position which are subject to strong syntactic restrictions (see Autenrieth, 2002). The respective properties of discourse markers and modal particles discussed so far are summarized in Table 3.1 for convenience. French bien is an adverb and a modal particle;1 Spanish bien, on the contrary, is an adverb and a discourse marker.
2. HOW
DO
DISCOURSE MARKERS ARISE? SPANISH BIEN
In the literature on the synchronic function of bien in contemporary Spanish, this discourse marker is often treated jointly with its near-synonym bueno (see Garce´s Go´mez, 1996; Martı´n
1
However, French has two discourse markers, bon and ben, of which bon in particular shares certain features with Spanish bien and bueno. Ben is sometimes considered a phonologically eroded variant of bien (Dostie, 2004). As its function is highly specific in Modern Standard French (see Hansen, 1995, 1998a), there is no functional overlap with the adverb bien. Therefore we do not consider ben a variant of adverbial bien.
46
Ulrich Detges and Richard Waltereit
Zorraquino and Portole´s, 1999). Both items have in common that they are used at places where discourse coherence is precarious in one way or another. Example (1) may serve as an illustration. Here, the speaker closes off a topic addressed by the previous speaker and embarks on a new one. A sudden change of topic of this kind could put at risk the coherence of the discourse in question. By using bien, the speaker signals that this switch takes place in a maximally cooperative fashion. This has been captured nicely by Serrano (1999: 121), who points out that both bien and bueno are devices for ‘‘negotiating’’ coherence. Apart from this, bien and bueno are used when the speaker disagrees with the previous speaker: (3) Bien as a marker of disagreement (Corpus Oral, 1992) S1: [E]l delito fiscal entre comillas es un privilegio de los ricos ¿no? o de [ . . . ] una clase social eh . . . con poder adquisitivo porque una familia media, una familia humilde, aunque quiera no puede cometer delito fiscal porque nunca defraudara´ ma´s de cinco millones a Hacienda. ‘Fiscal fraud, in quotation marks, is a privilege of the rich, isn’t it? or of a social class with purchasing power because an average family, a modest family, even if they wanted, cannot commit fiscal fraud because they would never cheat on more than five million [pesetas] to the tax authorities.’ S2: BIEN, pero de cinco millones, hay bastantes. Es decir, nos dice Hacienda, y creo que sabe de eso, que hay un treinta por ciento de defraudacio´n en este paı´s, un treinta por ciento [ . . . ]. ‘Well, but there a quite a few five-million fraud cases. In other words, the tax authorities tell us, and I think they know what they are talking about, that there is a rate of 31% of tax evasion in this country, 31%.’ Note that once again bien ensures a maximally cooperative exchange, thereby smoothing out the potential coherence gap. This analysis is in line with Hansen’s (1995, 1998a) account of French bon. According to her, bon is prototypically used to override the effect of some ‘‘undesirable’’ element in discourse or in the situation. This notion covers all sorts of phenomena, including risks for discourse coherence (see also Bre´mond, 2003). A third context type in which bien and bueno often appear are answers (for Spanish, see Fuentes Rodriguez, 1998: 55). This is the case when the incipient answer is in some way inconsistent with the presuppositions of the preceding question. Specifically, bien is used when the projected answer is going to be more complex than the question would have suggested. (4) Bien in answers (Corpus Oral, 1992) S1: [S]i [ . . . ] una persona ha tenido, por ejemplo una enfermedad, eh . . . hace tiempo, una hepatitis, una tuberculosis, etce´tera, y no se detecta en el ana´lisis correspondiente, de alguna manera . . . ¿esa persona que va a dar, o la que va a recibir pueden estar completamente tranquilas y seguras de que no se va a contagiar, esa otra persona?
Diachronic pathways and pragmatic strategies
47
‘If a person has had, for example, a disease, eh . . . some time ago, hepatitis, tuberculosis, and so on, and if this goes undetected in the respective analysis, in some way . . . , this person who will give [blood] or the one who will receive it, can they be entirely reassured and confident that they will not be infected, this other person?’ S2: BIEN, aunque no soy me´dico, sı´ debo contestar en alguna parte, aunque yo recomendarı´a que para estas cuestiones de ´ındole ma´s te´cnica, los propios servicios de hematologı´a hay me´dicos, personas perfectamente capacitadas, sin embargo, dentro de ese reconocimiento hay un cuestionario donde ante los riesgos, ante esas posibles enfermedades, son causa para declarar no apto a la persona. Ante la posible duda, mejor es que no done para evitar el posible contagio, por una enfermedad que no se desconoce y que puede filtrarse en los ana´lisis que se analizan. ‘Well, even though I am not a doctor, I do have to answer somehow, even though I would recommend that for these questions of more technical nature, the blood donor services themselves, there are doctors, perfectly qualified people, however, as part of the screening procedure there is a questionnaire where with respect to these risks, with respect to these possible diseases, they are grounds for declaring someone unsuitable. Considering this possible risk, it is better that this person should not give blood to avoid possible contagion, by a disease that is not unrecognized and that can be filtered out in the analyses that will be analyzed.’ S1 formulates a question whose answer requires some medical expertise. However S2, a political representative of the Spanish foundation of blood donors, does not feel suitably qualified to answer fully. Using bien, he conveys that he nevertheless accepts the question but is only going to talk about its administrative aspects. The question–answer pairs exemplify best how bien and bueno are used in the negotiation of coherence. On the one hand, the answering speaker explicitly accepts the question addressed to him. On the other hand, however, he does not provide an answer of the type projected in the question. S2 might as well have refuted the question or reformulated it in order to express his disagreement with the inapplicable expectations underlying it. Even though maximally informative and coherent, the latter alternative would be very costly in terms of encoding effort, and it might hamper the smooth progress of the exchange. Instead, the speaker using bien or bueno signals that she tacitly modifies the communicative task assigned to her by the previous question. A third, even less costly alternative would be to not express at all one’s disagreement with the question and to provide straightaway an answer which is judged most adequate to the interests expressed in the question. This alternative, however, would maximally put at risk the coherence of the unfolding discourse. Therefore, it seems appropriate only for minor mismatches between question and answer. This suggests that bueno and bien are mainly used for the repair of ‘‘medium-size’’ coherence issues, which should be recognized but need not be explicitly addressed.
48
Ulrich Detges and Richard Waltereit
In the discussed examples, bien is not used as an adverb. That is, it is not used to positively evaluate the foregoing chunk of discourse, as its lexical meaning would suggest. Now, what is the relation between the adverb bien and the homophonous discourse marker? In what follows, we shall sketch the diachronic evolution from adverb to discourse marker. Originally used exclusively as an adverb, bien is found in an unambiguous discourse-marker function only from the 18th century onwards. An early example in the CORDE corpus is (5). (5)
Bien as a discourse marker in 19th century Spanish (Larra, Doncel, 1834, CORDE) Ferrus: Estoy harto pagado con el honor de servirte-dijo el astuto juglar. ‘I’m abundantly paid with the honor to serve you, said the cunning minstrel.’ D. Enrique: BIEN, dejemos lisonjas que tu´ no crees ni yo tampoco; toma esas monedas ‘Well, let’s leave the flatteries that neither me nor you believe; take these coins.’
In this example, bien indicates that the second speaker – who apparently disapproves of the first speaker’s proposal – wishes to put aside the entire topic while at the same time continuing a cooperative exchange. The adverb bien ‘well’ is the source lexeme for a variety of linguistic expressions of different grammatical status. Many of these have in common that superficially positive evaluation is used for argumentation. First, in earlier periods of Spanish, bien is used as an interjection. In this function it indicates the speaker’s agreement with what has been said before (see (6)). This usage is already the outcome of a metonymic change which is based upon the inference that speakers normally evaluate favorably what they agree with. (6)
Adverbial bien in 16th century Spanish (Anonymous, Comedia Thebayda, 1500, CORDE) Menedemo: Todavı´a, Galterio, no quedo satisfecho de lo que me as dicho si no me dizes una cosa de que au´n me queda alguna dubda. ‘Still, Galterio, I am not satisfied with what you told me unless you tell me one thing about which I still have a doubt.’ Galterio: Sı´ hare´. Y por la encarnac- io´n del Verbo divino lo juro, y por la Vero´nica Santa de Jahe´n te lo prometo. ‘This I will do. And I swear by the Incarnation of the Divine Word, and by Holy Veronica of Jae´n I promise you.’ Menedemo: BIEN, abasta, y aun sin juramento te creyera. ‘Good, that’s enough, and I would have believed you without your oath.’
Second, adverbial bien may be used in concessive argumentation. After conceding the validity of an argument, the speaker goes on to put forward a much stronger counter-argument, as in (7).
Diachronic pathways and pragmatic strategies
(7)
49
Adverbial bien in concessive argumentation (Valde´s, Dia´logo de las cosas acaecidas en Roma, 1527–1529, CORDE) Latancio: Pues estando vosotros en pecado con vuestras mancebas, ¿no os parece que muy inominiosamente sois esclavos del pecado, y que os quita de´l el que procura que os case´is e viva´is honestamente con vuestras mujeres? ‘Thus, since you live in sin with your concubines, don’t you think that you are, very ignominiously, slaves of sin, and that what would you get out of this would be to get married and to live honourably with your wives?’ Arcidiano: BIEN, PERO ¿no vedes que parecerı´a mal que los cle´rigos se casasen, y perderı´an mucha de su auctoridad? ‘Right, but don’t you see that it would make a bad impression if clergymen were to marry, and that they would lose much of their authority?’
Clearly, the construction bien . . . pero ‘good . . . but’ is not the bridging context (i.e., a type of context that led from the old to the new interpretation) in which the change from adverb to discourse marker took place, seeing that the former, unlike the latter, crucially implies approval of the preceding stretch of discourse. However, as an instantiation of concessive argumentation, it is worthwhile to examine it in more detail, because concessive argumentation illustrates how a marker of positive evaluation such as bien can be used to express mild disagreement. According to classical assumptions of argumentation theory (Anscombre and Ducrot, 1977), the argumentational move p BUT q contains two arguments p and q, where p is an argument in favor of the conclusion r and q an argument against it. Uttering p BUT q, the speaker presents both arguments p and q as valid. However, she marks q (with its conclusion :r) as prevailing over p (with its conclusion r). In many examples of 16th century Spanish, p pero ‘but’ q is preceded, as in (7), by bien, which introduces the weaker argument p. The role of bien in this context is to explicitly acknowledge the validity of p without however canceling the effect of pero ‘but’, which makes q the stronger one of both arguments. Thus, in such contexts, bien, in spite of being a marker of positive evaluation, is used to express disagreement. Note that in contexts such as (7), bien functions, moreover, as an anaphoric device. The speaker of bien p pero q refers to an argument of her interlocutor and ascribes it the value of p. Thus, concessive argumentation is an example of negotiation in discourse: by uttering bien p pero q the speaker approves of an argument presented by her interlocutor, but only to ultimately refute it. This usage could at best explain the rise of bien as a concessive connective, as in si bien ‘even though’, but it does not explain the evolution of bien into a discourse marker. However, the argumentational move which paves bien’s way toward the role of a discourse marker is quite similar to concessive argumentation, as we will show in the following. A third diachronic outcome of argumentation with adverbial bien is the formula esta´ bien/bien esta´ ‘it is good’, which, again, is not the direct predecessor of bien as a discourse marker.
50
Ulrich Detges and Richard Waltereit
Originally, by virtue of its literal meaning, it expresses a positive evaluation of some state of affairs. Typically, this evaluation reflects that the speaker is satisfied (see (8)): (8)
Bien esta´ as a marker of positive evaluation (San Vicente Ferrer, Serm., 1411, CORDE) [ . . . ] aquellos nin˜os que mueren, si mueren bautizados, BIEN ESTA´ [ . . . ]. ‘Those children who die, if they die baptized, it’s good.’
A typical reason for positive evaluation, in turn, is the completion of a state of affairs to the speaker’s satisfaction. Thus, esta´ bien/bien esta´ ‘it is good’ expresses, as a result of a metonymic change, that the speaker considers an ongoing state of affairs as completed or even wants it to stop (see example (9)). This effect is doubled here by basta ‘enough’, which approximately conveys the same content as bien. (9)
Bien esta´ as a marker of completion (Anonymous, Relacio´n, c. 1541, CORDE) Y trujeron ocho mil hombres de los pueblos y conta´ronlos y mostra´ronselos a Guzma´n: ‘‘Basta; BIEN ESTA´.’’ ‘And they brought 8000 men from the villages, counted them and showed them to Guzma´n: ‘‘Enough, it is good.’’’
Esta´ bien/bien esta´ in the sense ‘enough, it is good’ can be an efficient rhetorical device in case the speaker wants to conceal her disagreement with some ongoing state of affairs to which she would like to put an end. In example (10), the second speaker (Constanza) invites Simo´n, a stutterer, to stop reading aloud, and does so in an extremely diplomatic fashion. The formula bien esta´, which can be understood as ‘it is good’, ‘it’s enough’, and ‘please stop’, fits this strategy perfectly. (10)
Bien esta´ as a rhetorical device for concealed disagreement (Lope de Vega, La nin˜ez del Padre Rojas, 1598, CORDE) Simo´n: Ave Marı´a, gra . . . gra . . . tia ple . . . plena Do . . . Domi . . . nus te . . . tecum, benedi. Constanza: No leas ma´s; BIEN ESTA´, porque el natural defecto no es culpa en ti. ‘Stop reading, it’s good, because that natural defect [i.e., your stutter] is not of your fault.’
In this sense, bien esta´ is sometimes used as part of an argumentational move bien esta´ p (ma´s bien) q ‘let’s stop doing p and (rather) start doing q’. This can be seen in the following passage, taken from the Paso de Leno y Sulco, su amo, sobre el rato´n, a 16th century farce. Sulco, a rich cattle owner, has found his servant Leno sleeping in the haystack, but does not recognize him. Leno wants to make him believe that he is not Leno but a kind of exotic mouse imported from America. At some point of their absurd exchange, Sulco cuts off the conversation and orders his other servants to seize Leno and to tie him up – this brusque change of activity is introduced by bien esta´.
Diachronic pathways and pragmatic strategies
(11)
51
Bien esta´ announcing change of activity (Lope de Rue´da, Pasos, 1535, CORDE) Sulco: Si tan grandes son como vos los ratones de vuestra tierra, los gatos que los cac- aren, ¿de que´ taman˜o pueden ser? ‘If the mice of your country are as big as you, what might be the size of the cats which hunt them?’ Leno: Sen˜or, sera´n de quinze leguas de largo. ‘Sir, they will be 15 miles long.’ Sulco: ¿Y de ancho? ‘And how wide?’ Leno: Veynte y dos. ‘Twenty-two.’ Sulco: ¿Co´mo es possible ser ma´s anchos que largos? ‘How is it possible to be bigger in width than in length?’ Leno: Por que son hechos ancho por largo. ‘Because they are made width by length [and not length by width].’ Sulco: ¿Y que´ hazı´ades vos en mi pagiza? ‘And what were you doing in my haystack?’ Leno: Sen˜or, entre´me huyendo de un cabo de guayta. ‘Sir, I came in because I was fleeing from a sentinel officer.’ Sulco: Ora BIEN ESTA´. A´tenle al brocal de aquel pozo y no le den de comer bocado hasta que venga quien le conozca. ‘It’s good/that’s enough now. Tie him to this well and don’t give him anything to eat until someone comes who knows him.’
Accordingly, apart from positive evaluation, i.e., its literal meaning, the formula bien esta´ has developed two main contexts of use, namely implicit disagreement and change of activity. Turning to bien as a discourse marker, it would appear that it has traveled an analogous pathway of diachronic change as bien esta´. The crucial difference between the two however is that bien esta´ may refer to any joint human activity, whereas bien as a discourse marker is specialized in one specific kind of activity, namely the joint construction of discourse (see (5), repeated for convenience): (5)
Bien as a discourse marker in 19th century Spanish (Larra, Doncel, 1834, CORDE) Ferrus: Estoy harto pagado con el honor de servirte-dijo el astuto juglar. ‘I’m abundantly paid with the honor to serve you, said the cunning minstrel.’ D. Enrique: BIEN, dejemos lisonjas que tu´ no crees ni yo tampoco; toma esas monedas. ‘Well, let’s leave the flatteries that neither me nor you believe; take these coins.’
Like bien esta´ as a marker of change of activity, bien as a discourse marker is the diachronic outcome of an argumentation relating two instructions, namely, first, the instruction to end an
52
Ulrich Detges and Richard Waltereit
‘‘old’’ activity and, second, the instruction to begin a ‘‘new’’ one. This explains why, as a discourse marker, it has a two-place structure, relating two chunks of discourse. More properties of the discourse marker bien can be explained from the diachronically underlying argumentation. Thus, like bien esta´ in (11), bien as a discourse marker can express concealed disagreement (see (3), (5)), while at the same time reflecting the speaker’s willingness to pursue the discourse with a new topic. Thus, the diachronic path sketched here explains the synchronic intuition that bien serves to ‘‘negotiate’’ coherence. The comparison between the development of bien esta´ and bien brings to light another important property of discourse markers. Recall that bien esta´ is not a discourse marker, but a formulaic device which serves to coordinate human activities of any kind. The purpose of bien as a discourse marker, however, is to coordinate the joint construal of discourse. Strictly speaking, discourse markers are just a subset of the routines human agents have at their disposal for the coordination of their joint activities. Thus, it is not surprising that the diachronic evolution of a discourse marker should follow the same pathway as such a routine. The same reasoning seems to apply to other discourse markers such as now as in (12), which serves to introduce a new topic into the discourse. (12)
Now as a discourse marker (Jowett, tr., Thucydides Peloponnesian Wars, 1881, OED) Now the Acharnians are famous for their skill in slinging.
Its source lexeme, the time adverbial now, can likewise be used to announce new referents or new events, introducing an abrupt change in the sequence of activities projected by the participants:2 (13)
Temporal now coordinating unspecific human activities (www.bbc.co.uk/ouch/podcast/ transcripts/ouch_podcast6_transcript.rtf ) And now, ladies and gentlemen, listen to this.
Further cases in point are look and listen, which as discourse markers draw the attention of the hearer to the upcoming stretch of discourse (see (14), (16)). As free imperatives, however, they can be used to make the listeners aware of all kinds of suitable linguistic or extra-linguistic stimuli (cf. Brinton, 2001; Waltereit, 2002); see (16), (18): (14) (15)
Look as a discourse marker (Sime, Hunters Point, 1973, OED) Look, we don’t have to sit here. We could go down to the beach. Look as a free imperative (Hurston, Mules & Men, 1935, OED) ‘Looka here, folkses,’ Jim Presley exclaimed. ‘Wese a half hour behind schedule.’
2 For Spanish ahora ‘now’ see Silva-Corvala´n (2001: 214–237), for French maintenant ‘now’ see Saussure (2008), for Old French or ‘now’, see Ollier (2000).
Diachronic pathways and pragmatic strategies
(16) (17)
53
Listen as a discourse marker (www.buffalobeast.com/80/shred.htm) Umm, listen man; I need a little time here. Listen as a free imperative Hey, listen to this!
Another case in point is French tiens ‘look’, which derives from the imperative form of tenir ‘to hold’. (18)
Tiens as a discourse marker il pleut! ‘Look, it rains!’
TIENS,
Among other things, our analysis explains why imperative verb forms are an important diachronic source of discourse markers (Dostie, 2004). After all, imperatives are conventionally specialized in coordinating human activities. Conversation is a locally managed human activity (Sacks et al., 1974). After every move, the participants have to decide anew upon the question: ‘‘What are we going to do next?’’ There is a constant requirement to negotiate and re-negotiate the following move. The successful outcome of this activity is discourse coherence. Discourse markers arise because lexemes are used for argumentational procedures which are helpful in those negotiations. Discourse markers of the type examined in this section are the routinized residue of negotiations about the next move.
3. FRENCH BIEN – THE RISE
OF A
MODAL PARTICLE
As shown in (19) and (20), the French modal particle bien is used in assertions and questions: (19)
(20)
French bien in assertions (Hansen, 1998b: 111) C’est BIEN la premie`re fois que c- a m’arrive! ‘That IS the first time that this happens to me!’ French bien in questions ( ¼ (2)) Vous avez BIEN rec- u mon message? ‘You did receive my message, didn’t you?’
In (19), bien serves to reinforce the assertion. As has been shown by Hansen (1998b), bien is polyphonous. According to the theory of polyphony (Ducrot, 1984; for a more recent comprehensive account see Iten, 1999), some lexical items encode ‘‘controversies’’ between various ‘‘voices’’ (e´nonciateurs, E), each of which may be associated with a conversation participant, i.e., the actual speaker herself (locuteur, L) and others (e.g., the hearer). In this model, an assertion with bien evokes two e´nonciateurs, E1 and E2. One of these, E2, who is associated with the speaker L, denies the viewpoint of another e´nonciateur (E1).
54
Ulrich Detges and Richard Waltereit
(21)
Polyphonous bien a. (E1: Ce N’est PAS la premie`re fois que vous eˆtes en retard.) ‘This is not the first time you are late!’ b. L ( ¼ E2): C’est BIEN la premie`re fois que je suis en retard. ‘This IS the first time I am late!’
As can be seen in (21), the first e´nonciateur E1 evoked by bien always negates the propositional content of the utterance (21a), before E2, associated with the speaker’s point of view (21b), refutes E1’s negation. Moreover, bien systematically implies that in this controversy E2 ‘‘wins out’’ over E1. In short, bien signals the speaker’s refutation of a negation coming from another discourse participant (typically, but not necessarily, the hearer). Negative assertions of the type (21a) may, but need not, occur in the preceding context of the utterance (21b). In the normal case, they are simply evoked or assumed. In this sense, assertions with bien are the mirror image of simple sentence negation (Hansen, 1998b: 121), which also systematically implicates an opposite point of view (Givo´n, 1979: 107) and is therefore another instance of polyphony (see Ducrot, 1984: 214–223). (22)
Polyphonous sentence negation a. (E1: C’est la premie`re fois que je suis en retard.) ‘This is the first time I am late.’ b. L ( ¼ E2): Ce n’est pas la premie`re fois que vous eˆtes en retard. ‘This is not the first time you are late!’
Givo´n (1995: 114) proposes a communicative re-definition of modality. According to him, categories of epistemic modality such as assertion, negation, irrealis assertion and presupposition not only make reference to the degree of subjective certainty on the part of the speaker. Above all, they anticipate the hearer’s reaction to the validity of the speaker’s utterance. In other words, they express the degree to which the speaker assumes her utterance to be (un)controversial in the eyes of the hearer. Given this interactional view of modality, our argumentation makes it clear what is ‘‘modal’’ about modal particles. The modal particle bien makes reference to strong counterexpectation on the part of the hearer, just as negation does. It is therefore an instantiation of communicative modality (see Hansen, 1998b). It seems to us that this applies to modal particles in general. They crucially refer to participants’ stance toward speech acts (Waltereit, 2001). To that extent, modal particles are inherently polyphonous. For the sake of completeness, it should be mentioned that bien is used not only in assertions, but also in questions: (23)
Bien in questions C’est BIEN la premie`re fois que vous eˆtes en retard? ‘Is it actually the first time you are late?’
Diachronic pathways and pragmatic strategies
55
In questions, bien is typically used by the speaker (L ¼ E2) to further inquire about a previously raised issue. By virtue of its polyphonous structure, it evokes a foregoing assertion by the hearer (E1) to the effect that the corresponding proposition is true. This assertion can either be an actual utterance or it can be inferable from the context. (24)
a. [E1: C’est la premie`re fois que je suis en retard.] ‘This is the first time I’m late.’ b. L ¼ E2: C’est BIEN la premie`re fois que vous eˆtes en retard? ‘Is it actually the first time you are late?’
Taken together, both uses of bien ((22) and (24)) show that modal particles are polyphonous items specialized for certain types of speech acts. In Ducrot’s (1984) theory, polyphony is a conventional property of linguistic items. That is, it belongs to the level of langue, not of parole. In the following we will show that diachronically this property emerges from the conventionalization of effects in dialogic exchanges. The modal particle bien, which has been in use since Old French, goes back to adverbial bien ‘well’. Like in Modern French, the latter served to express a positive evaluation of a state of affairs: (25)
Bien ‘well’ as an adverb (Chevalier de La Charrette 455, BFM) Li mangiers fu BIEN atornez. ‘The meal was well prepared.’
However it is very unlikely that the adverbial bien ‘well’ is the direct diachronic predecessor of the homophonous modal particle. Rather, the latter must go back to an intermediate stage, namely the degree adverb bien ‘at lot, very much, to a large extent’, as exemplified in (26). (26)
Ne sevent pas ne ne sont BIEN certain. (Ami et Amile 3119, c. 1200, BFM) ‘They do not know and aren’t very sure.’
We assume that the bridging contexts, where adverbial bien ‘a lot’ changed into the modal particle, were of the type (27). In these contexts, the degree adverb is used to argue against strong counter-expectation on the part of the hearer: (27)
From adverb to modal particle: bridging context (Robert de Clari, La Conqueˆte de Constantinople, CVI, 31, BFM) Et mesires Pierres respondi: ‘‘Ba!’’, fist il, ‘‘de n’ave´s vous oı¨ comment Troies le grant fu destruite ne par quel tor? – Ba ouil!’’, fisent li Blak et li Commain, ‘‘nous l’avons BIEN oı¨ dire.’’ ‘And Mylord Pierre answered: ‘‘Ba’’, he said, ‘‘haven’t you heard about how Troy the great was destroyed and in which way this happened?’’ ‘‘Of course’’, said Blak and Commain, ‘‘we did hear a lot about it.’’’
56
Ulrich Detges and Richard Waltereit
From today’s perspective, bien in (27) could either be the adverb ‘a lot, to a large extent’ or the homophonous modal particle. Construed as an adverb, it is used in (27) as part of a strategy of scalar argumentation. It serves to effectively counter a denial. In the dialogical context (27), L2 (Blak and Commain) refute the foregoing negation by L1 (mesires Pierre).3 Contrary to L1’s claim that the proposition p does not obtain, they contend that p not only does hold, but that it even holds TO A LARGE EXTENT. The argumentational move underlying the rise of the modal particle bien invokes a dialogical exchange of the kind given in (28). (28)
L1: p is not the case. L2: p is the case TO A
LARGE EXTENT!
Example (27) is typical of Old French uses of bien insofar as the latter often appears together with verbs of knowledge (savoir ‘to know’) or verbs of perception (oir ‘to hear’, entendre ‘to listen,’ and voir ‘to see’) expressing states of knowledge, as nous l’avons bien oı¨ dire ‘we did hear a lot about it’. In these contexts, bien underscores the speaker’s belief that his knowledge is valid despite an expectation to the contrary on the part of the hearer. Note that the hearer’s stance does not necessarily need to be openly expressed. In example (29), adverbial bien is used to deny a counter-expectation which is only implicit. It is reflected in the king’s reply, especially in part following mez ‘but’, which could be interpreted to the effect that the king does not believe his counselor and consequently does not take his complaints seriously. This in turn means that (28) can also appear in the less explicit form (28u). (29)
(28u)
Bien with presumed counter-expectation (JoinvCB, 00748) L’en raconte du roy Phelippe, [ . . . ] que une foiz li dit un de ses conseillers que moult de torfaiz li fesoient ceulz de sainte Esglise [ . . . ]. Et le bon roy respondi que il le creoit BIEN, mez il regardoit les bonte ´ s et les courtoisies que Dieu li avoit faites, si vouloit miex lesser aler de son droit que avoir contens a lagent de sainte Esglise. ‘People tell of King Philip that once a counsellor of his told him that the Church was doing him [i.e., Philip] much harm. And the good king replied that he believed this well/to a large extent, but that he focused on the good and noble things God had given to him, and that he preferred not to insist on his right rather than to get in conflict with the people of the Holy Church.’ [L1: p is not the case.] L2: p is the case TO A LARGE EXTENT!
The scheme (28), (28u), is an instance of scalar argumentation, invoking a scale as in (30). This means that the relation of bien and negation is not simply a syntagmatic one (use of bien
3 In (27) we are however not dealing with a plain negation but with a rhetorical question oriented toward a negative answer.
Diachronic pathways and pragmatic strategies
57
prototypically follows negation contexts). Bien and negation are also in paradigmatic relation, as they represent opposing endpoints on a scale of validity of p. (30)
a. p is not the case b. p is the case c. p is the case
TO A LARGE EXTENT!
Scalar argumentation means that (30c) p is the case TO A LARGE EXTENT is invoked as an argument in favor of the conclusion that (30b) p is the case. This is done against strong counterexpectation, i.e., against the hearer’s belief that (30a) p is not the case is true. This argumentational move is given in (31). In (31) the speaker says that p is the case TO A LARGE EXTENT but he really means to say that p REALLY is the case, despite strong expectation to the contrary. (31)
p is the case ‘thing said’
TO A LARGE EXTENT
- p REALLY is the case (despite expectation to the contrary) ‘thing meant’
Before the change from adverb to modal particle, the ‘thing meant’ in (31), i.e., the conclusion that p REALLY is the case is a mere inference. The shift from bien ‘to a large extent’ to a modal particle takes place if this inference becomes conventionalized by frequent usage, and turns into the new meaning of bien. By the same token, the original effect of scalar argumentation will disappear altogether. This is shown in example (32). While in (27) and (29) bien is ambiguous between an interpretation as adverb or modal particle, we can be sure to deal with a modal particle in (32): (32)
Old French bien as a modal particle (Ami et Amile 1724–1729, c. 1200, BFM) – Ber, pran ma fille par la toie merci [ . . . ]. ‘Baron, take my daughter, for goodness sake!’ – Non ferai certez, ce li respont Amis, car dans Hardre´z fu BIEN de cest paı¨s, asse´z i a et parens et couzins, en traı¨son m’ avroient tost ocis. ‘This I will surely not do, replied Ami, because Lord Hardre´ did come from this land, there are a lot of his relatives and cousins around, they would soon manage to kill me insidiously.’
The difference between bien in (27) and (29) on the one hand and (32) on the other is that (32) excludes a gradable interpretation. There are no degrees of ‘‘coming from this land’’. One is either
58
Ulrich Detges and Richard Waltereit
from a given land or not. Hence, (32) is not suitable for a scalar argumentation of the type (28). In (32) bien simply indicates that some state of affairs REALLY is the case despite expectations to the contrary. If this analysis is correct, then we can describe the relationship between the diachronic motivation of the change and the synchronic function of modal particles. Diachronically, the modal particle bien arises from a dispute about the validity of a certain proposition p. The synchronic residue of this is a polyphonous element which encodes speakers’ assumptions about the hearer’s attitude toward the validity of an assertion. In Old French, bien also appears in questions, as it does in Modern French (see above (23)). An example is (33): (33)
Old French bien in questions (Joinville, La Vie de Saint Louis, 394, c. 1300, BFM) Et je li diz: ‘Ne savez vous pas BIEN que se vous mourie´s en ce point, que vous seriez seriez damne´ et iriez en enfer?’ Et il dit: ‘Oy¨l’. ‘And I told him: ‘‘Know you not really that if you die at this moment, you will be damned and go to hell?’’ And he said: ‘‘Yes’’.’
The function of bien is the same as in assertive contexts: what is known WELL/A LOT (against an expectation to the contrary) is REALLY known. Bien in questions is inherently related to bien in assertions insofar as a question with bien could systematically be answered with an assertion containing bien. However, in actual Old French texts this is rather rare, as the normal item for replying to enquiries of this type is Old French certes ‘sure’, which is stronger than simple oui ‘yes’ (see (34)). (34)
Old French bien and certes (Jehan de Paris 68, BFM) ‘‘Comment,’’ dit la pucelle, ‘‘vous semble il BIEN que ung roy de France pouroit BIEN autant faire comme cestuy?’’ ‘‘Certes, ma doulce seur, je croy que ouy [ . . . ]’’ ‘‘‘What’’, said the girl, ‘‘does it really seem to you that a king of France could really do as much as this one?’’ ‘‘Sure, my dear sister, I think yes [ . . . ]’’’.
Even though the function of modal bien in questions is the same as in assertive contexts, it would seem that the change from adverbial bien ‘a lot’ to the modal particle occurred later in questions than it did in assertions (Schulze, 1888: 87). What can we learn from this change about the rise of modal particles? First, this change is metonymic (see Koch, 1999, 2001). A form with the meaning ‘a lot, to a large degree’ is used to reinforce an assertive speech act in contexts of counter-expectation. Second, the modal particle emerges from a stereotyped argumentation which originally is dialogical in nature (see (28) and (28u)). Change of this kind may account for the polyphonous semantics of modal particles in general.
Diachronic pathways and pragmatic strategies
59
4. CONCLUSION As we have shown, both discourse markers and modal particles arise through metonymic semantic change. Furthermore, both emerge from dialogical argumentation patterns. The basic difference between both diachronic pathways lies in the nature of the respective argumentation. Discourse markers arise in contexts where speakers negotiate their further verbal interaction (‘‘What are we going to do next?’’). The argumentational patterns used here are the same as for the negotiation of non-verbal interactions. Modal particles, on the contrary, arise from stereotypical argumentational moves negotiating common ground (‘‘What do I believe that you believe concerning the felicity of my speech act?’’). The notion that the meaning of pragmatic markers may be specifically related to argumentation and the negotiation of viewpoints has been previously studied (e.g., Schwenter, 2000; Schwenter and Traugott, 2000). However to our knowledge, it has not been sufficiently appreciated that the contemporary meaning of these items may arise as the by-product of argumentational routines for which their diachronic forerunners were actually used. Our model provides motivation for Traugott’s (1999), Traugott and Dasher’s (2002: 187, 225), and Brinton and Traugott’s (2005) subjectification hypothesis, i.e., the assumption that metatextual meanings tend to be recruited from propositional ones. But in addition, our model shows how and why different types of metatextual meaning arise in diachrony. In past research, both the evolution of discourse markers and the emergence of modal particles have been analyzed within theoretical frameworks which aim at broad generalizations about semantic change. Thus, they have both been labeled as instances of grammaticalization or ‘‘pragmaticalization’’ (Silva-Corvala´n, 2001; Autenrieth, 2002; Wegener, 2002; Diewald, 2006). While the diachronic changes discussed in this paper do not contradict any of these assumptions, it seems clear to us that there are relevant differences which these labels fail to capture. Specifically, the latter are far too general to understand the respective synchronic function of both discourse markers and modal particles from a diachronic perspective.
REFERENCES Anscombre, J.-C. and O. Ducrot (1977). ‘‘Deux ‘mais’ en franc- ais?’’ Lingua 43: 23–40. Autenrieth, T. (2002). Heterosemie und Grammatikalisierung bei Modalpartikeln. Eine synchrone und diachrone Studie anhand von eben, halt, ech(er)t, einfach, schlicht und glatt. Tu¨bingen: Niemeyer. Bre´mond, C. (2003). ‘‘Bon, moteur d’action, moteur du discours’’. Travaux Interdisciplinaires du Laboratoire Parole et Langage 22: 65–84. Brinton, L. (1996). Pragmatic markers in English: Grammaticalization and discourse functions. Berlin: Mouton de Gruyter.
60
Ulrich Detges and Richard Waltereit
Brinton, L. (2001). ‘‘From matrix clause to pragmatic marker: The history of look-forms’’. Journal of Historical Pragmatics 2 (2): 177–199. Brinton, L. and E. C. Traugott (2005). Lexicalization and language change. Cambridge: Cambridge University Press. Diewald, G. (2006). ‘‘Discourse particles and modal particles as grammatical elements’’, in K. Fischer (ed.), Approaches to discourse particles. Oxford: Elsevier, 403–426. Dostie, G. (2004). Pragmaticalisation et marqueurs discursifs. Analyse se´mantique et traitement lexicographique. Brussels: De Boeck Duculot. Ducrot, O. (1984). Le dire et le dit. Paris: Minuit. Fuentes Rodriguez, C. (1998). La sintaxis de los relacionantes supraoracionales. Madrid: Arcos. Garce´s Go´mez, . M. P. (1996). ‘‘Los marcadores discursivos en espan˜ol’’, in A. Gil and C. Schmitt (eds.), Koha¨sion, Koha¨renz, Modalita¨t in Texten romanischer Sprachen. Bonn: Romanistischer Verlag, 126–147. Givo´n, T. (1979). On understanding grammar. New York: Academic Press. Givo´n, T. (1995). Functionalism and grammar. Amsterdam: Benjamins. Hansen, M.-B. M. (1995). ‘‘Marqueurs me´tadiscursifs en franc- ais parle´: l’exemple de bon et de ben’’. Le franc- ais moderne 63: 20–41. Hansen, M.-B. M. (1998). The function of discourse particles. A study with special reference to spoken standard French. Amsterdam: Benjamins. Hansen, M.-B. M. (1998). ‘‘La grammaticalisation de l’interaction ou Pour une approche polyse´mique de l’adverbe bien’’. Revue de Se´mantique et Pragmatique 4: 111–138. Iten, C. (1999). ‘‘The relevance of argumentation theory’’. UCL Working Papers in Linguistics 11: 41–81. Koch, P. (1999). ‘‘Frame and contiguity: On the cognitive bases of metonymy and certain types of word formation’’, in G. Radden and K.-U. Panther (eds.), Metonymy in language and thought. Amsterdam: Benjamins, 139–167. Koch, P. (2001). ‘‘Metonymy: Unity in diversity’’. Journal of Historical Pragmatics 2: 201–244. Martı´n Zorraquino, M. A. and J. Portole´s (1999). ‘‘Los marcadores del discurso’’, in I. Bosque and V. Demonte (eds.), Grama´tica descriptiva de la lengua espan˜ola. Madrid: Espasa Calpe, 4051–4213. Ollier, M.-L. (2000). Linguistique de l’e´nonciation d’une langue morte. Analyse de Or avez vos folie dite. In: M. -L. Ollier (ed.), La forme du sens. Textes narratifs des XIIe et XIIIe sie`cles. E´tudes litte´raires et linguistiques. Orle´ans: E´ditions Paradigme, 433–459. Sacks, H., E. A. Schegloff and G. Jefferson (1974). ‘‘A simplest systematics for the organization of turn-taking in conversation’’. Language 50: 696–735. Saussure, L. de (2008). ‘‘Maintenant: pre´sent cognitif et enrichissement pragmatique’’, in M. Vuillaume (ed.), Ici et maintenant. Cahiers Chronos 20. Amsterdam: Rodopi, 53–76. Schulze, A. (1888). Der altfranzo¨sische direkte Fragesatz. Ein Beitrag zur Syntax des Franzo¨sischen. Leipzig: Hirzel. Schwenter, S. (2000). ‘‘Viewpoints and polysemy: Linking adversative and causal meanings of discourse markers’’, in B. Kortmann and E. Couper-Kuhlen (eds.), Cause-condition-contrast-concession. Berlin: Mouton de Gruyter, 257–281. Schwenter, S. and E. C. Traugott (2000). ‘‘Invoking scalarity: The development of ‘in fact’’’. Journal of Historical Pragmatics 1: 7–25. Serrano, M. J. (1999). ‘‘Bueno como marcador discursive de inicio de turno y contraposicio´n: estudio sociolingu¨´ıstico’’. International Journal of the Sociology of Language 140: 115–133. Silva-Corvala´n, C. (2001). Sociolingu¨´ıstica y pragma´tica del espan˜ol. Georgetown: Georgetown University Press.
Diachronic pathways and pragmatic strategies
61
Traugott, E. C. (1997). ‘‘The role of the development of discourse markers in a theory of grammaticalization’’. Paper presented at ICHL XII, Manchester 1995; Version of 11/97, available at: http://www.stanford. edu/Btraugott/papers/discourse.pdf Traugott, E. C. (1999). ‘‘The role of pragmatics in a theory of semantic change’’, in J. Verschueren (ed.), Pragmatics in 1998: Selected Papers from the 6th International Pragmatics Conference, Vol. 2. International Pragmatics Association, Antwerp, 93–102. Traugott, E. C. and R. Dasher (2002). Regularity in semantic change. Cambridge: Cambridge University Press. Waltereit, R. (2001). ‘‘Modal particles and their functional equivalents: A speech-act-theoretic approach’’. Journal of Pragmatics 33: 1391–1417. Waltereit, R. (2002). ‘‘Imperatives, interruption in conversation and the rise of discourse particles: A study of Italian guarda’’. Linguistics 40: 987–1010. Wegener, H. (2002). ‘‘The evolution of the German modal particle denn’’, in I. Wischer and G. Diewald (eds.), New reflections on grammaticalization. Amsterdam: Benjamins, 379–394.
FURTHER REFERENCES
Corpora BFM ¼ Base de Franc- ais Me´die´val. 74 textes inte´graux d’ancien et de moyen franc- ais, rassemble´s par Ce´line Guillot, Serge Heiden, Alexey Lavrentiev et Christiane Marchello-Nizia [bfm.ens-lsh.fr]. CORDE ¼ Real Academia Espan˜ola. Corpus Diacro´nico del Espan˜ol. [corpus.rae.es/cordenet.html]. Corpus Oral ¼ Francisco A. Marcos-Marı´n et al. 1992. Corpus de Referencia de la Lengua Espan˜ola Contempora´nea: Corpus Oral Peninsular. Universidad Auto´noma, Madrid [www.lllf.uam.es/Bfmarcos/ informes/corpus/corpulee.html]. DAVIES ¼ Marc Davies. Corpus del Espan˜ol. [www.corpusdelespanol.org]. OED ¼ John Simpson et al. (2009), Oxford English Dictionary, Oxford University Press, Oxford [dictionary.oed.com].
Current Trends in Diachronic Semantics and Pragmatics Edited by Maj-Britt Mosegaard Hansen and Jacqueline Visconti r 2009 by Emerald Group Publishing Limited. All rights reserved.
4 CONTEXT SENSITIVE CHANGES: THE DEVELOPMENT OF THE AFFIRMATIVE MARKERS GODT ‘GOOD’ AND VEL ‘WELL’ IN DANISH Eva Skafte Jensen
1. PRELIMINARIES In Modern Danish the word godt ‘good’ in certain contexts has the function of what I shall call an affirmative marker: In a statement such as the one in (1), godt is used to affirm that the nonnegative counterpart to a conceivable negative statement is indeed the intended: (1) Jeg har godt hørt det ‘I have good1 heard it’ Certainly, I have heard it A paraphrase trying to capture the effects of godt in this utterance might run as follows: ‘Someone might think that I did not hear it, but that’s a wrong supposition, because, indeed, I did hear it’. This article outlines how the word godt came to develop this function in 19th century Danish. I shall claim that the development may be analyzed as ‘‘the conventionalizing of conversational implicature’’, a formulation originally coined by Traugott and Ko¨nig (1991; cf. also Traugott, 1999; Traugott and Dasher, 2005; Hansen and Waltereit, 2006). 1
The lexical item godt does not translate into a single word or expression in English. As will become apparent, its meaning is dependent on syntactic function, word order and other factors. For this reason, in all examples godt is simply glossed as ‘good’, even though ‘well’ in some cases might be more in tune with Modern English semantics. To make up for this, in addition to the gloss, a more idiomatic translation is given wherever necessary and possible.
64
Eva Skafte Jensen
A second claim is that the changes depicted in this article are multifactorial, i.e., that the affirmative function of godt is the result of more than one single influential factor. Factors of particular relevance to the development of the affirmative marker godt are the position of the word in the sentence (word order) and the pragmatic–semantic context of the utterances in which godt occurs. Section 2 gives an account of the close relationship between morphosyntax, word order, and meaning of words like godt in Danish. Section 3 gives an outline of how the development of affirmative godt came about, and how semantic–pragmatic features as well as word order were influential in this process. A third claim is that having an affirmative marker is not new to the Danish lexico-grammatical system. The case of how vel ‘well’ developed a function as affirmative marker in the 15th century is briefly presented in Section 4. As a matter of interest to my first claim mentioned above, the development of affirmative vel is similar to the one of affirmative godt. This, too, may be analyzed as ‘‘the conventionalizing of conversational implicature’’.
2. GODT AS AFFIRMATIVE MARKER
IN
MODERN DANISH
As one of my claims is that the development of godt into an affirmative marker is due to several factors, one of them being that the syntactic function and the meaning of the word may be ambiguous, and that the ambiguities and possible disambiguation mechanisms are linked to the word order in Danish, this entire Section 2 concerns the relevant features of morphosyntax and word order in Modern Danish. 2.1. Morphosyntactic features in Modern Danish The affirmative marker godt is a specialized function of the adverbial godt ‘good’. The adverbial godt stems from the adjective god ‘good’. (2) (3)
god-Ø (common gender) en god-Ø appelsin ‘a goodcommon orange’
god-t (neuter gender) et god-t æble ‘a goodneuter apple’
Adding a neuter-t to an adjective is one of the most productive ways of forming adverbials in Danish as in (4)–(6). The adding of a neuter-t to an adjective in Danish resembles the adding of the adverbial -ly to an adjective in English. (4) (5)
den/det smager god-t ‘itcommon/itneuter tastes goodneuter ’ hun løber hurtig-t ‘she runs fastneuter ’
Context sensitive changes
(6)
65
hun synger god-t ‘she sings goodneuter ’
Thus, in addition to the function as affirmative marker (cf. (1)), godt appears as an ordinary manner adverbial in sentences such as (4)–(6), and it also has a function as an intensifier of the predicate (degree word, cf. Bolinger, 1972) as in (7) (Jensen, 2001).2 (7)
Han er godt dum ‘He is good stupid’ He is rather stupid
As a manner adverbial, godt translates into well. As an intensifier it translates into very (much), or rather. Sometimes it is impossible in a given sentence to determine whether godt functions as a manner adverbial or as an intensifying adverbial. This is illustrated by (8). (8)
Jeg tænkte godt over sagen ‘I thought good over the-case’ (8a) I thought about the case in a good manner (8b) I thought about the case quite a lot With no context to support one interpretation over the other, both translations of (8a) and (8b) are fully valid.3 Of specific relevance to the development of the affirmative function shown in (1) above, are the function as manner adverbial (cf. (4)–(6)), the function of intensifier (cf. (7)), and the fact that in some cases the function and meaning of the lexical item godt may be ambiguous (cf. (8)). The ambiguities illustrated by (8) are partly linked to the feature of word order in Danish. The manners in which ambiguous examples like the one in (8) may be analyzed, and how they may be disambiguated are shown below in Section 2.3. Before that, however, a few prerequisites on the word order possibilities of the different functions of godt are given in Section 2.2. 2.2. Word order, semantics, and syntactic function As in other languages, lexical items in Danish hold a certain meaning potential rather than one fixed meaning (Bartsch, 1984; Durst-Andersen and Herslund, 1996; Togeby, 2003; Traugott and Dasher, 2005; Hansen and Strudsholm, 2008). Several factors determine to what extent the rather
2
The same might be said for some of the uses of the French word bien ‘well’. According to Hansen, under certain circumstances, this word may be said to function ‘‘comme marqueur de renforcement’’ (Hansen, 1998: 114). 3 In (8), godt may even function as an affirmative marker. A fair translation might be: Yes, I did think about the case.
66
Eva Skafte Jensen
open meaning potential is narrowed down into a more unambiguous meaning, and one of these factors is word order. As a regular adverbial of manner modifying the predicate of the sentence, godt has a certain topological flexibility,4 as it may be placed after the finite verb and the negation (cf. (9)), or it may be fronted as in (10): (9)
(10)
Hun synger ikke godt ‘She sings not good’ She doesn’t sing well Godt synger hun ikke ‘Good sings she not’ She doesn’t sing well
As an intensifier godt is more restricted as regards word order. If the word functions as a qualification of the contents of a single word (typically an evaluative adjective) as in (11), it always appears to the immediate left of this word. If it functions as a qualification of the predicate as a whole, it appears to the immediate right of the finite verb (12): (11)
(12)
Det har gjort mig godt gal ‘It has made me good angry’ It has made me very angry Hun kører godt til ‘She drives good to’ She drives very fast
The word order of affirmative godt is even more restricted. It is fixed to a position in the so called ‘central field’ of the sentence (cf. Section 2.3 below). In declarative sentences without subjectverb-inversion,5 this position is to the immediate right of the finite verb, (13)–(15): (13)
4
Jeg har godt hørt det ‘I have good heard it’ Indeed, I’ve heard it
Under certain circumstances, manner adverbials may even be placed to the very far right of the sentence, i.e., several places to the right of the finite verb, as in han kom kørende hurtigt lit.: ‘he came driving fast’. 5 The statement that affirmative godt is fixed positionally to the central field holds in all sentence types. However, in declarative sentences with subject-verb-inversion, in sentences with an unstressed nominal constituent enclitically attached to the verb, in subordinate clauses and in interrogative clauses, due to the word order rules specifically pertaining to these sentence types, this position may not be immediately after the finite verb as was the case in (13)–(15), cf. det havde han godt hørt lit.: ‘that had he good heard’; han hørte-odet godt lit.: ‘he heard-oit good’; . . . at han godt har hørt det lit.: ‘ . . . that he good have heard it’; har du godt hørt det? lit.: ‘have you good heard it’.
Context sensitive changes
(14)
(15)
67
Jeg sa˚ godt hun kom for sent ‘I saw good she came to late’ Yes, I saw that she was late Du ma˚ godt la˚ne min cykel ‘You may good borrow my bike’ Sure, you may borrow my bike
Affirmative godt is not mobile at all. Whereas the manner adverbial godt could move about, and let itself be fronted, for instance, fronting of the godt in a sentence like (15) is not an option: (16) simply does not make sense, and therefore I have refrained from supplying the sentence with an idiomatic translation. (16) *Godt ma˚ du la˚ne min cykel ‘Good may you borrow my bike’ To sum up, as a manner adverbial godt has a certain flexibility in terms of word order, as an intensifier godt also has a few options as regards word order, but functioning as an affirmative marker the position of godt is not optional at all, but quite fixed. It only occurs in the so called ‘central field’. In the next section the characteristics of this central field are given, and special attention is paid to the possible ambiguities as to meaning, function and word order of a word like godt. These possible ambiguities are partly accountable for the development of godt into an affirmative marker.
2.3. Word order in Modern Danish, characteristics of the central field In Danish, the different positions in the sentence have specialized functions (Diderichsen, 1946; Jensen, 2002b; Hansen and Heltoft, to appear). Of interest to the matter at hand, are the characteristics of the position where affirmative godt is placed. This particular position is also the slot for the negation ikke ‘not’ and other negational operators such as kun ‘only’ etc., and for sentence adverbials with epistemic and interpersonal meanings such as sikkert ‘probably’, ma˚ske ‘maybe’, forha˚bentlig ‘hopefully’, and many more.6 In accordance with Danish grammar tradition
6 Cohesive adverbials like dog ‘although, however’, alligevel ‘nevertheless’, altsa˚ ‘thus, ergo’ also appear here, as does a certain group of highly context sensitive discourse markers and modal particles like altsa˚, nu, jo, skam. The lexical items in the latter group do not correspond 1:1 to lexical items in English. An attempt to explain their meaning in English is made by Davidsen-Nielsen (1996).
68
Eva Skafte Jensen
I call this position the central field (Diderichsen, 1946; Hansen, 1977; Allan et al., 1995; Hansen and Heltoft, to appear): subject finite verb (17)
(18)
(19)
(20)
(21)
CENTRAL FIELD
POST FIELD
sentence adverbials and negation, etc.
non-finite verbs and other constituents, incl. manner adverbials hørt det heard it’
Jeg har ikke ‘I have not I haven’t heard it Vi har kun ‘We have only We have only received two applications Du ma˚ sikkert ikke ‘You may probably not She probably won’t let you borrow her bike Hun laver ma˚ske ‘She makes maybe Maybe, she’ll cook for us Det er forha˚bentlig ‘It is hopefully It’s possible, I hope
fa˚et to ansøgninger got two applications’ la˚ne hendes cykel borrow her bike’ mad til os food for us’ muligt possible’
Important to the argument that word order is an influential factor as regards the development of godt into a marker of affirmation, is the fact that in some sentences it cannot be determined whether a word such as godt is, in fact, placed in the central field or further to the right in of sentence. This may be the case of very short sentences with just one verb. In a sentence such as jeg forsta˚r det godt (lit.: ‘I understand it good’), godt may be perceived to be placed either in the central field, (22), or further to the right in the sentence, (23). In constructions with modal verbs or other auxiliaries, such as the aspectual use of have ‘have’, i.e., constructions of expanded predicates, this ambiguity does not arise, (24)–(25):
CENTRAL FIELD
(22) (23) (24) (25)
Jeg forsta˚r det yes, I understand it Jeg forsta˚r det I understand it well Jeg har yes, I’ve understood it Jeg har I’ve understood it well
post field
godt godt godt
forsta˚et det forsta˚et det godt
Context sensitive changes
69
These synchronic aspects of the intricate interplay between meaning, function, ambiguity, disambiguation and of word order in Modern Danish are examined in Jensen (1997, 2000a, b, 2002a, b, 2007). In Section 3 the diachronic aspects of how word order may come to be one of the influential factors in the development of godt is shown.
3. THE REINTERPRETATION
OF GODT AS A
MARKER
OF
AFFIRMATION
The development of godt into a marker of affirmation seems to have taken place in the 19th century as a reinterpretation of the manner adverbial/intensifier.7 The feature of possibility is an important part of the semantics of affirmative godt, and the earliest examples of godt functioning unambiguously as an affirmative marker, are found in utterances where this feature is overt and an integral part of the semantics of the predicate, as in (26)–(28) (taken from dictionaries of dialects and colloquial speech, Feilberg, 1886–1893; Kristiansen, 1908), with modal verbs like kan ‘can’, ma˚ ‘may’, and vil ‘will’. (26)
(27)
(28)
det kan godt være ‘it can good be’ it’s quite possible det ma˚ du godt ‘it may you good’ you may do it jeg vil godt komme ‘I will good come’ yes, I’ll come
In other types of utterances, where the feature of possibility is more covert and of a pragmatic nature rather than semantic, unquestionable examples of affirmative godt is not found until several decades later, like the one in (29), where godt does not function as a manner adverbial, nor as an intensifier. (29)
Jeg havde godt lagt mærke til at en klynge damer havde ligget oppe i marehalmen og set pa˚ os ‘I had good laid mark to that a cluster ladies had lain up in the-lyme-grass and looked at us’ I had, in fact, noticed that a group of ladies had been watching us from behind the lyme grass up above (Hjortø, Dengang man var ung, 1908: 125)
7 In earlier works I have touched on this subject (Jensen, 1997, 2002a), but the outline given in this article on the actual process of development of godt is in fact the result of new research.
70
Eva Skafte Jensen
In the literature of the early 20th century, many examples can be found (cf. ODS, the adjective god ‘good’). This means that the reinterpretation of the lexical item godt must have taken place before the latter part of the 19th century, but that the use of affirmative godt in contexts of less explicit semantic possibility than the ones conveyed by kan ‘can’, ma˚ ‘may’, and vil ‘will’, probably does not happen until several decades later. The reinterpretation itself most likely took place in what is now commonly known as bridging contexts (cf. Hansen, 2008: 62–63ff ). Going back in time, to the early part of the 19th century, we find some interesting examples where godt may have been used as an intensifier or a manner adverbial, but perceived as something else (namely a marker of affirmation). (30) and (31) are both taken from a book of hymns written by the famous preacher and moralist Grundtvig. In (30) and (31) the full stanza is given, but only the important line (30:1), (31:7) containing godt is supplied with a word-for-word-gloss. (30)
(31)
1. Du veed det godt, Guds Menighed! ‘You know it good, Gods devotees!’ 2. Det ene tjener til din Fred, 3. At i dit Hjærte og din Mund 4. Er Troens Ord i allen Stund. (Grundtvig, nr. 123, 1875 [1832]) 1. O, Kristelighed! 2. Du skjænker vort Hjærte, 3. hvad Verden ej veed; 4. Hvad svagt vi kun skimte, 5. mens Øjet er blaat, 6. Det lever dog i os, 7. det føle vi godt; ‘it feel we good’ 8. Mit Land, siger Livet, 9. er Himmel og Jord, 10. Hvor Kjærlighed bor! (Grundtvig, nr. 214, 1875 [1832])
You know it well, devotees of God! This alone gives you Peace, That in your Heart and in your Mouth The Word of Faith exists forevermore. Oh, Christianity! You grant our Heart, what the World does not know; What vaguely we only glimpse, while the Eye is blue, That lives within us, we feel that well/strongly; My Country, says Life, is Heaven and Earth, Where Love lives!
In these examples from Grundtvig, godt probably is intended as an intensifier or a manner adverbial. Whether the word does in fact function as one or the other in (30) or (31) is impossible to determine (cf. (8) above) and not important to the point at hand. Important, however, is the nature of the verbs within the scope of godt in the two examples, as they concern subjective beliefs and emotions, in (30) vide ‘know’, in (31) føle ‘feel’. These verbs are rather different to the verbs løbe ‘run’ and synge ‘sing’ in (5) and (6) above, in that they belong to another semantic domain altogether. Verbs such as vide ‘know’ and føle ‘feel’
Context sensitive changes
71
are subjective expressions, whereas verbs such as løbe ‘run’ and synge ‘sing’ without any modalization prototypically enter into statements open to intersubjective debate. Another important factor beyond the semantics of the verbs is the contextual pragmatics of the utterances containing godt. In (30) Du veed det godt ‘you know it good’ is of an imploring or beseeching nature. Grundtvig is reminding (or maybe even commanding) the congregation of the proper way of obtaining peace of mind through faith. And in much the same way he calls upon the congregation to feel in a certain manner in (31). One might say that there were no need for such appeals were there no doubt as to whether the congregation were on the right track, so to speak, as to how to be properly faithful. Thus the intensifier/manner adverbial godt appears in surroundings of conceivable doubt and uncertainty. The appearance of godt in examples such as these may then very well be perceived as markers of positive value in opposition to a (potentially) negated utterance. Thus the godt may be perceived as the counterpart of the negation ikke ‘not’. (30)u Du veed det ikke (31)u vi føle det ikke
‘you know it not’ ‘we feel it not’
In this manner the intensifier/manner adverbial godt may give rise to the implicature that, indeed, there might be reason to think that someone (e.g., Du ‘you’ in (30), vi ‘we’ in (31)) does not know or feel what needs to be known and felt, in much the same way as the negation in the famous statement the whale is not a fish gives rise to the implicature that someone does indeed believe the non-negated statement, i.e., that the whale is a fish. In the latter part of the 19th century we find examples like the ones in (32) and (33): (32)
(33)
Du har ingen Fortrolighed til mig, Karl. Du ved dog godt, at jeg kan tie (E. Brandes, 1882) ‘you have no confidence to me, Karl. You know dogmodalparticle good, that I can keepquiet’ You have no confidence in me, Karl. You know quite well that I can hold my tongue Disse Linjer gør Indtryk af at Shakespeare har set Corregio’s berømte Billed Jupiter og Io. Det er ogsaa godt muligt ‘These Lines give Impression of that Shakespeare have seen Corregio’s famous Picture Jupiter and Io. It is also good possible’ These lines give the impression that Shakespeare may have seen the famous painting Jupiter and Io by Corregio. And indeed, that is quite possible (G. Brandes, 1895: 135)
As was the case in (30) and (31), the context in both of the examples are within the realm of conceivable doubt and possibility. In (32) a character in a play tries to ascertain her trustworthiness to a reluctant other character, and as in (30) the verb in (32) is vide ‘know’. In
72
Eva Skafte Jensen
(33) the essayist confirms the possibility of a certain scenario: that Shakespeare could indeed have seen a certain painting by Corregio. As in (30)–(31), the appearance of the word godt in (32)–(33) may give rise to the implicature that someone might believe the negative counterpart of the utterance in which godt occurs: (30)u (31)u (32)u (33)u
Du veed det godt vi føle det godt Du ved dog godt Det er ogsaa godt muligt
(impl. (impl. (impl. (impl.
someone someone someone someone
thinks thinks thinks thinks
that that that that
du veed det ikke) vi føle det ikke) du ved ikke) det er ikke muligt)
And as already mentioned, this kind of implicature resembles the kind found in negated utterances, only with the opposite value of polarity: (34)
the whale is not a fish
(impl. someone thinks that
the whale is a fish)
The person who often hears examples like the ones in (30)–(33) may then conclude that the intensifier/manner adverbial godt is in fact an affirmative marker godt, and thus the conversational implicature generated from examples like (30)–(33) above becomes part of the conventional meaning of the word godt’s lexical meaning and functional potential (Traugott and Ko¨nig, 1991). A scenario like this suggests that the reinterpretation happens on the part of the hearer, rather than the speaker. As pointed out by Nielsen (2005: 180f ), even if the speaker has a specific intention in uttering a certain statement, the inferences are worked out by the hearer, and may therefore – within reason – be beyond the control of the speaker. Grundtvig, E. Brandes and G. Brandes may all have had the intention of qualifying (by intensity or by manner) the predicate of the sentence by godt, or they may innovatively have used godt as a marker of affirmation. But the kind of implicature suggested in (30)u–(33)u may have happened independently of speaker intention. In this kind of change, then, the Invited Inferencing Theory of Semantic Change, as formulated by Traugott and Dasher (2005), may not apply. 3.1. Word order as influential factor The conventionalizing of the conversational implicature shown in the previous section is supported by the word order. As was shown earlier, (17)–(21), the central field – among other things, cf. footnote 6 – contains the negation ikke ‘not’ and other negational operators such as kun ‘only’, etc., and sentence adverbials with modal and interpersonal contents such as sikkert ‘probably’, ma˚ske ‘maybe’, forha˚bentlig ‘hopefully’, etc. The central field in itself, then, can be said to be a position for – at least – negation and modality. In sentences such as the ones in (30)– (33) godt could easily be perceived as being placed in this central field, and as an effect of this, the negational and interpersonal/modal properties of the position per se may influence the semantic
Context sensitive changes
73
properties of the lexical item, godt, when placed in that exact position. Thus godt comes to form a paradigmatic relation with the negation.
(30/32) (31) (33) (30/32)u (31)u
subject finite verb du ved det det føle vi det er du ved det vi føle det
CENTRAL FIELD
POST FIELD
sentence adverbials and negation godt godt ogsa˚ godt ikke ikke
non-finite verbs and other constituents, incl. manner adverbials
muligt
Taking this into account, for the contemporary language user to perceive the intensifier or manner adverbial godt as a negational operator with some modal and interpersonal properties is not really that strange, in fact it makes perfect sense. 3.2. The paradigm of negational ikke ‘not’, affirmative godt, and Ø The changes of the functional use of godt in the 19th century has led to the current situation in Modern Danish: We now have a small paradigm revolving around polarity with the lexical members ikke ‘not’ and godt. Besides these two overt lexical members there is a third member of the paradigm, represented by the option Ø.8 Traditionally, there is some debate as to whether negation or non-negation is more basic philosophically speaking (Horn, 2001: 9–10, 60f). From a linguistic point of view, however, it seems reasonable that a statement containing a negation is less basic than a non-negated statement. In Jensen (2007), within the framework of la ScaPoLine,9 I suggest that statements containing affirmative godt in Modern Danish be analyzed into three distinct points of view. The analysis builds on the observation made by polyphonists such as Ducrot and others, that a negated statement in itself contains two points of view (two ‘voices’). The one point of view is represented by the propositional contents of the statement without the negation, the second point of view is represented by the negation of the propositional contents. Following this analysis, the non-negated statement is then more basic than the negated statement. In my analysis of affirmative godt in Modern Danish, I suggest that affirmative godt in fact adds yet another point of view to the two points of view already present in the
8
The core paradigm contains the three members mentioned: affirmative godt, the negation ikke, and Ø. On occasion, however, other lexical items may be used as a marker of affirmation, e.g., sagtens in utterances such as der er sagtens plads lit.: ‘there is sagtens room’, i.e. certainly, there’s plenty of room. 9 La the´orie Scandinave de la Polyphonie Lingustique, developed by Henning Nølke, based on the thoughts on polyphony by Oswald Ducrot (cf. Nølke et al., 2004).
74
Eva Skafte Jensen
negated statement. By using affirmative godt, the speaker activates, so to speak, all three points of view at once, see (29), (35)–(36), with (36) as the most basic statement and (29) as the most complex statement.10 (29) (35) (36)
Jeg havde godt lagt mærke til at en klynge damer. . . I had, in fact, noticed that a group of ladies . . . Jeg havde ikke lagt mærke til at en klynge damer. . . I didn’t notice that a group of ladies . . . Jeg havde Ø lagt mærke til at en klynge damer. . . I had noticed that a group of ladies . . .
(impl. someone thinks that:) (impl. someone thinks that:)
Affirmative godt apparently has some semantic constraints worth mentioning when attempting to describe the properties of the paradigm of which affirmative godt is part. For instance, it is often found with modal verbs of possibility, (37)–(38), but it does not combine with modal verbs of necessity, (39): (37)
(38)
(39)
Han kan godt være kommet uden at jeg har lagt mærke til det (epistemic possibility) ‘He can good be come without that I have laid notice to it’ He may very well have arrived without my noticing it Du ma˚ godt la˚ne min cykel (deontic possibility; permission) ‘You may good borrow my bike’ You’re welcome to borrow my bike *Du skal godt købe ind inden 5 (deontic necessity; obligation) ‘You must good buy in before five’
This suggests that affirmative godt does not cancel just any kind of negation, but, rather, that it functions as a negational operator within a certain domain, and this certain domain is the classic modal domain of possibility, whereas it does not apply to the domain of necessity. Thus, the semantics of affirmative godt include (at least) two important features, namely affirmation (i.e., cancellation of a negation) and modality. As the modal properties in question lie within the domain of possibility, certain contextual effects can be observed. In addition to the semantic–pragmatic effects displayed by the utterances in (37)–(38), where godt reinforces the semantics of (non-negative) possibility, there are some other notable effects too. The most obvious is the one seen in argumentative texts. Often affirmative godt co-occurs with words and phrases with adversative and even concessive
10 Readers of Detges and Waltereit will notice that my analysis differs from their analysis of the French modal particle bien ‘well’. In their analysis, the three options: (1) bien, (2) negation, and (3) Ø, form a continuum with Ø in the middle, flanked by bien on the positive side of the scale, and the negation on the negative side (Detges and Waltereit, 2007: 76).
Context sensitive changes
75
meanings such as men ‘but’, selvom ‘although’, dog ‘however, ikke desto mindre ‘nevertheless’, etc. (40) shows an example: (40)
‘‘Jeg kunne og kan ikke se, hvad der skulle være i vejen for at psykisk udviklingshæmmede kan have kunstnerisk talent. Det er mennesker, der har problemer med at læse og fungere socialt, men derfor kan de godt være kreative og have kunstnerisk talent’’, siger Poul Welle. ‘‘‘I could and can not see, what there should be in the-way for that mentally retardees can have artistic talent. It is humans who have problems with to read and function socially, but therefore can they good be creative and have artistic talent’’, says Poul Welle’ I could not, and still can’t see why the mentally disabled cannot be artistically talented. We’re talking about people with limited academic and social skills, but this does not necessarily mean that they lack creative and artistic skills’’, says Poul Welle (Jyllandsposten, 5/4 1998)
The speaker in (40) argues against the assumption that dysphatic people should be without artistic talent, and in doing so, he – among other things – makes use of the affirmative godt. For affirmative godt to appear in argumentative texts like the one in (40) is quite common.11
4. VEL,
A
MIDDLE DANISH PREDECESSOR
As stated in the preliminary remarks in Section 1, Danes have made use of affirmative markers for at least 500 years. Affirmative vel ‘well’ was used as the affirmative marker in Danish until it was ousted by affirmative godt. Of particular interest to the topic of this article, the adverbial vel apparently went through a process similar to the one described above concerning godt (Jensen, 1997, 2002a).12 In the case of vel, the function as affirmative marker seems to have developed during the 15th century. In the manuscript called Mariager Legendeha˚ndskrift dating back to the late 15th century, we find several examples (in the examples vel is spelled wæll): (41)
11
GVtz brudh sancta katherina fornam wæll, at thet war ihesu syn kæreste brudgommes wiliæ, at hwn skulle stundom omganges meth menneske ‘Gods bride saint catherine sensed well, that it was jesus her dearest bridegroom’s will, that she should sometimes associate with people’
Hansen (1998) and Detges and Waltereit (2007) have observed some similar points in their analysis of the French modal particle bien ‘good’. They, too, find that in some contexts, the word functions as marker of refutation (Hansen, 1998: 119; Detges and Waltereit, 2007). 12 As a matter of interest, the English word well went through a similar process as well, cf. Finnel, 1989. The grammatical results, however, especially as regards word order, differs from the ones in Danish.
76
Eva Skafte Jensen
(42)
The bride of God, St. Catherine, strongly sensed (or: sensed strongly) that it was the wish of Jesus, her beloved groom, that she should occasionally associate with people (Mariager Legende-Haandskrift, 398, 15) Tha wisthe paffwen wæll, at Sanctus Jeronimus war fulkommen i latinæ maall, gretske oc jødskæ ‘Then knew the-pope well, that Saint Jerome was perfect in latin language, greek and jewish’ Then the pope knew for certain that St. Jerome was perfect in the languages Latin, Greek and Jewish (Mariager Legende-Haandskrift, 371, 17)
The factors influencing the reinterpretation of vel are by and large the same as the ones applying to the development of godt in the 19th century. For one thing, it should be noted that as was the case by the development of godt in the 19th century, the semantics of the predicates in (41) and (42) are of a rather subjective nature, fornam ‘sensed’ and wisthe ‘knew’, and this is likely to have some influence on the semantics of vel. More importantly, however, are the pragmatics at work in examples like the ones in (41) and (42). A person listening to examples such as (41) and (42) may infer that someone might think that St. Catherine did not quite know how Jesus wanted her to behave, and that someone might think that the Pope did not really know whether the linguistic skills of St. Jerome were in order. Therefore, examples like (41) and (42) give rise to the implicature that there might be reason to believe the negative counterpart of the affirmative statement, and this conversational implicature, in turn, has a bearing on the conventional meaning of vel. After a while the presence of vel is perceived as a means to cancel or negate the possible negative counterpart, and thus again we find an example of the conventionalizing of a conversational implicature, and we find that the word vel forms a paradigmatic relation with the negation. In support of this idea of vel entering into a paradigm with the negation, I’ll point to a small survey I conducted of the combinational options of modal verbs and the affirmative marker vel in the time around 1500 (Jensen, 1997). In a small corpus I had found 225 attested manifestations of the modal verb thore ‘dare’. Of these 225 utterances 11 contained the affirmative marker vel; almost all of the remaining utterances contained some kind of negation, either in the form of a ‘pure’ negation ey/ikke ‘not’, or some negated pronoun like inthet ‘nothing’. In Modern Danish, vel no longer carries the affirmative function of earlier times. In the right type of constructions it still contains a certain amount of appeal to the addressee, and it does activate more than one point of view (polyphonically speaking) (Krylova, 2005), but a sentence like (43), which is constructionally parallel to (41) and (42) above, definitely sounds odd, and at best archaic, at worst misunderstood as an epistemic marker of some uncertainty. As a consequence I have refrained from supplying (43) with an idiomatic translation.
Context sensitive changes
(43)
77
Hun forstod vel hvad han ønskede hun skulle gøre ‘She understood well what he wished she should do’
Until the latter part of the 19th century, vel was the dominant affirmative marker of Danish. The emergence of affirmative godt, however, eventually led to the ousting affirmative vel. Before the final disappearance of affirmative vel, there must have been a period of coexistence. Exactly for how long this period has lasted, still remains to be uncovered in detail.
5. CONCLUSIONS I have made several claims in this article. (1) Danes have been using lexicalized affirmation markers for more than 500 years. From Middle Danish up to Modern Danish the lexical item in use as an affirmative marker was vel. Later on, beginning sometime in the 19th century at the time newly developed affirmative godt overtakes this function, and affirmative vel is ousted. (2) Lexically, both words vel and godt have expanded their meaning potential (A - AB), and as a consequence of that, first vel and then godt have added yet another function to their domain. But besides the expansion lexically and functionally, one might also describe the changes in terms of restriction: the affirmative marker is positionally fixed as opposed to other meanings and functions, and it is a part of a small grammatic–semantical paradigm with only a few other members. These are features often mentioned in works describing processes of grammaticalization (see Lehmann, 1985; Hopper and Traugott, 1993). (3) The reinterpretation process in the case of both vel and godt proves to be context sensitive in more senses than one. First, the subjective semantics of the verbs in the sentences in which vel and godt occur affect the interpretation of vel and godt. Second, the properties of the word order position in which the word may be perceived to be placed (the central field) may also affect the meaning potential of the word. Finally, the pragmatic surroundings of the examples with vel and godt may also influence the way the word is perceived by way of implicature. Thus, both cases (vel and godt) may be described as a process whereby conversational implicatures turn into conventional meaning, i.e., an integral part of the potential meaning of the word.
ACKNOWLEDGMENTS I wish to thank the research group Sprog og Rationalitet (SPRAT) at Roskilde University, and the participants at the workshop on Diachronic Semantics and Pragmatics at the 18th International Conference on Historical Linguistics (ICHL) for discussing earlier versions of this work with me.
78
Eva Skafte Jensen
REFERENCES Allan, R., Ph. Holmes and T. Lundskær-Nielsen (1995). Danish, a comprehensive grammar. London: Routledge. Bartsch, R. (1984). ‘‘Norms, tolerance, lexical change, and context-dependence of meaning’’. Journal of Pragmatics 8: 367–393. Bolinger, D. (1972). Degree words. The Hague: Mouton. Brandes, E. (1882). Gyngende Grund. Skuespil i fire Akter. Kjøbenhavn: P. G. Philipsens Forlag. Brandes, G. (1895). ‘‘William Shakespeare’’, Samlede Skrifter, Vol. 8. Kjøbenhavn: Gyldendalske Boghandels Forlag & Græbes Bogtrykkeri, 1901. Davidsen-Nielsen, N. (1996). ‘‘Discourse particles in Danish’’, in E. Engberg-Pedersen, M. Fortescue, P. Heltoft, L. Heltoft and L. F. Jakobsen (eds.), Content, expression and structure – Studies in Danish functional grammar. Amsterdam: John Benjamins, 283–314. Detges, U. and R. Waltereit (2007). ‘‘Different functions, different histories. Modal particles and discourse markers from a diachronic point of view’’. Catalan Journal of Linguistics 6: 61–80. Diderichsen, P. (1946). Elementær dansk grammatik. 3rd Edition, 9th Imprint. København: Gyldendal. 1987. Durst-Andersen, P. and M. Herslund (1996). ‘‘The syntax of Danish verbs. Lexical and syntactic transitivity’’, in E. Engberg-Pedersen, M. Fortescue, P. Harder, L. Heltoft and L. F. Jakobsen (eds.), Content, expression and structure. Studies in Danish functional grammar. Amsterdam: John Benjamins, 65–102. Feilberg, H. F. (1886–1893). Bidrag til en Ordbog over Jyske Almuema˚l. Vol. I. Kjøbenhavn: UJdS & Thieles Bogtrykkeri. Finnel, A. (1989). ‘‘Well now and then’’. Journal of Pragmatics 13: 653–656. Grundtvig, N. F. S. (1875). Salmer og aandelige Sange. Vol. IV. Kjøbenhavn: Kristen-Livet i Salme-Sang, Karl Schønberg & Bianco Lunos Bogtrykkeri. Hansen, E. (1977). Dæmonernes Port. 3rd Edition. København: Reitzel. 1984. Hansen, E. and L. Heltoft (to appear). Grammatik over det danske sprog. København: Gyldendal. Hansen, M.-B. M. (1998). ‘‘La grammaticalisation de l’interaction ou Pour une approche polyse´mique de l’adverbe bien’’. Revue de Se´mantique et Pragmatique 4: 111–138. Hansen, M.-B. M. (2008). Particles at the semantics/pragmatics interface: Synchronic and diachronic issues. A study with special reference to the French phasal adverbs (Current Research in the Semantics/ Pragmatics Interface, Vol. 19). Amsterdam: Elsevier. Hansen, M.-B. M. and E. Strudsholm (2008). ‘‘The semantics of particles: Advantages of a contrastive and panchronic approach. A study of the polysemy of French de´ja` and Italian gia`’’. Linguistics 46 (3): 471–505. Hansen, M.-B. M. and R. Waltereit (2006). ‘‘GCI theory and language change’’. Acta Linguistica Hafniensia 38: 235–268. Hjortø, K. (1908). Dengang man var ung. København. Hopper, P. J. and E. C. Traugott (1993). Grammaticalization. Cambridge: Cambridge University Press. Horn, L. R. (2001). A natural history of negation. Chicago: Chicago University Press. Jensen, E. S. (1997). ‘‘Godt og gerne – modalitet og polaritet’’, in P. Widell and M. Kunøe (eds.), Møde om ˚ rhus: Aarhus Universitet, 154–164. udforskningen af dansk sprog. Vol. 6. A Jensen, E. S. (2000a). ‘‘Sætningsadverbialer og topologi med udgangspunkt i de konnektive adverbialer’’, in J. Nørga˚rd-Sørensen, P. Durst-Andersen, L. Jansen, B. L. Jensen, and J. Pedersen (eds.), Ny forskning i grammatik. Fællespublikation 7. RASK Supplement Vol. 11. Odense: Odense universitetsforlag, 141–154. Jensen, E. S. (2000b). Danske sætningsadverbialer og topologi i diakron belysning, Ph.D. thesis, University of Copenhagen.
Context sensitive changes
79
Jensen, E. S. (2001). ‘‘Om subjektivitet som motivation for sproglig forandring’’, in P. Jarvad, F. Gregersen, L. Heltoft, J. Lund, and O. Togeby (eds.), Sproglige a˚bninger. E som Erik, H som 70. København: Festskrift til Erik Hansen, Reitzel, 194–204. Jensen, E. S. (2002a). ‘‘Hvordan bliver et sætningsadverbial til et sætningsadverbial?’’, in K. Kristensen (ed.), Studier i nordisk 2000-2001. København: Foredrag og a˚rsberetning, Selskab for Nordisk Filologi, 31–50. Jensen, E. S. (2002). ‘‘On the development of Danish sentence adverbials’’, in H. Jansen, P. Polito, L. Schøsler and E. Strudsholm (eds.), L’infinito & oltre. Omaggio a Gunver Skytte. Odense: Odense University Press, 463–480. Jensen, E. S. (2007). ‘‘Negationer med kompleks semantik i polyfonisk belysning’’, in R. Therkelsen, ˚ rhus: Aarhus N.M. Andersen and H. Nølke (eds.), Sproglig polyfoni. Tekster om Bachtin og ScaPoLine. A Universitetsforlag, 261–282. Kristiansen, V. (1908). Ordbog over Gadesproget og saakaldt daglig Tale. 2nd Edition. København: Hagerup. Krylova, E. (2005). ‘‘Epistemisk polyfoni i danske modalpartikler’’. Sproglig polyfoni. Arbejdspapirer. Vol. 4. Roskilde: Sprogligt polyfoninetværk & Institut for Sprog og Kultur, Roskilde Universitetscenter, 75–88. Lehmann, Chr. (1985). ‘‘Grammaticalization: Synchronic variation and diachronic change’’. Lingua e Stile 20 (30): 303–318. Mariager Legende-Haandskrift in the manuscript Gl. kgl. Saml. 1586 41 (1488), in G. Knudsen (ed.), Samfund til Udgivelse af gammel nordisk Litteratur. København: S. L. Møllers Bogtrykkeri, 1917–1930. Nielsen, N. M. (2005). Counter argument. In defence of common sense. Roskilde: Roskilde Universitetsforlag. Nølke, H., C. Nore´n and K. Fløttum (2004). ScaPoLine. La The´orie Scandinave de la Polyphonie Linguistique. Paris: Kime´. ODS ¼ Ordbog over det danske Sprog (1918–1954). V. Dahlerup et al. (eds.). København: Dansk Sprog- og Litteraturselskab & Gyldendal. Togeby, O. (2003). Fungerer denne sætning? Funktionel dansk sproglære. København: Gad. Traugott, E. C. and E. Ko¨nig (1991). ‘‘The semantics-pragmatics of grammaticalization revisited’’, in E.C. Traugott and B. Heine (eds.), Approaches to grammaticalization Vols. I–II. Amsterdam: John Benjamins Publishing Company, 189–218. Traugott, E. C. and R. B. Dasher (2005). Regularity in semantic change (Cambridge studies in linguistics, Vol. 97). Cambridge: Cambridge University Press. Traugott, E. C. (1999). ‘‘The role of pragmatics in semantic change’’, in J. Verschueren (ed.), Pragmatics in 1998. Selected papers from the 6th International Pragmatics Conference, Vol. 2. Antwerp: International Pragmatics Association, 93–102.
Current Trends in Diachronic Semantics and Pragmatics Edited by Maj-Britt Mosegaard Hansen and Jacqueline Visconti r 2009 by Emerald Group Publishing Limited. All rights reserved.
5 PROCATALEPSIS AND THE ETYMOLOGY HEDGING AND BOOSTING PARTICLES
OF
Kate Beeching
1. INTRODUCTION In her groundbreaking 1990 book From etymology to pragmatics. Metaphorical and cultural aspects of semantic structure, Eve Sweetser demonstrates that a cognitive approach to meaning can account for regularities in three important linguistic domains: polysemous relationships, lexical semantic change and pragmatic ambiguity. She argues that there are clear metaphorical or metonymic relationships between Meaning 1 (M1) and Meaning 2 (M2) in each of these domains and that these metaphorical relationships are universal and based in human cognition. What I wish to argue in this chapter is that, in a more restricted set of cases, another rhetorical figure, procatalepsis, may be invoked to describe the M1/M2 relationship and that this explains the somewhat paradoxical hedging-cum-boosting quality of pragmatic particles such as quand meˆme, though, but and aber. Examples 1–4, drawn, in the main, from corpora of contemporary spoken language and typical of everyday usage, illustrate the hedging/boosting qualities of these four particles: (1)
(2) (3)
C’est une ville qui a quand meˆme un cine´ma. ‘It is a town which has all the same a cinema.’ It’s a town which does (after all) have a cinema. (Beeching Corpus, Interview 4, line 35) Nice house though isn’t it? BNC spoken corpus I was teaching the wee man to ride his bike the day. It’s good to see him cycling but. Overheard example, Dumbarton, Scotland
82
Kate Beeching
(4)
‘‘Du hast aber Mumm!’’ Rief sie bewundernd aus. (HC01.283) ‘‘You have but guts!’’ shouted she admiringly out. FSF: ‘‘Well, you are spunky!’’ she exclaimed admiringly. (Translation of Scott Fitzgerald’s The Beautiful and the Damned.)
Most of these examples (or similar cases) will be discussed in more detail further down. Suffice it for the moment to say that each of these particles has a corresponding concessive or adversative subordinating or coordinating conjunctival form, each retains the trace of its contrastive forebear, each has, to differing degrees, a curiously emphatic yet apologetic, boosting and hedging quality. The business of this chapter is to shed light on the cognitive process whereby a concessive or adversative conjunction (M1) undergoes a semantic and grammatical change to become a hedging and boosting particle (M2) – and what might be the motivation for such a usage.
2. RHETORICAL EFFECTS Metaphor and metonymy ‘‘are said to constitute an interpretive use, with the sentences being used to convey a relationship of resemblance between the proposition expressed and the thought(s) it is intended to convey’’ (Kempson, 2001: 414). In my analysis, procatalepsis has a similar rhetorical effect, which is, however, quite particular to itself. In ancient Greek rhetoric, procatalepsis is a figure which allows a speaker to concede certain arguments and thus strengthen his or her main argument. In ordinary conversation, there is a similar procataleptic strategy which seems to work in two stages; first, drawing on the core, propositional, meanings of words used in interaction, the speaker concedes that there may be an opposing view – the assertiveness of the speech act can be hedged and a potential criticism disarmed. Second, by downtoning the force of an assertion and conceding part of the argument, the speaker is paradoxically able to be more convincing and boost the argument. In the (possibly fictive) suggestion that there is an opposing view, the speaker attends to her interlocutor’s face needs, comes across as likeable and can thus simultaneously hedge and boost the proposition. The paper takes as a starting-point Kempson’s (2001: 414) notion that ‘‘all word meaning is but a set of procedures . . . [which] suggests that any word is but the input to the construction of some novel ‘‘ad hoc’’ concept specific to that utterance’’. In ordinary everyday interaction, speakers frequently employ language in ways which are designed to protect the faces of both speaker and hearer (Brown and Levinson, 1987) – and it is therefore unsurprising that words with semantic cores to do with contrast/concession are recruited to serve face-preserving needs. Contrast and concession allow the speaker to suggest that another view is possible. Metonymic processes (routinization resulting from repeated contiguity in the syntagmatic chain) thence confirm the M2, which, instead of hedging, performs the opposite: boosting. This process of
Procatalepsis and the etymology of hedging and boosting particles
83
pragmaticalization1 occurs unevenly. It occurs, however, in words with similar semantic cores across typologically different languages, such as English, German, French, Spanish, Chinese and Japanese, suggesting that it derives from universal cognitive reflexes as well as universal politeness principles. The pragmaticalization process which leads from M1 to M2, from a lexical or grammatical item to a semantically relatively bleached pragmatic particle, is generally considered to occur through the semanticization of ‘‘invited inferences’’ (Traugott and Dasher, 2002) and to be a gradual and unidirectional process. Though this chapter focuses mainly on the cognitive ‘actuation’ phase of meaning change (the mechanism which yields the pragmatic implicature or invited inference), rather than the ‘propagation’ phase (whereby what could be a nonce-usage generalizes across a language community), some attention is paid to the fine-grained tracking of uses across spoken and written text-genres, an approach to propagation recommended by Hansen (2008: 288) as ‘‘worthy of attention’’.
3. PROCATALEPSIS
AND
CONCESSION
Procatalepsis – or prolepsis – is a classical ‘figure of thought’ described by Quintilian (1921– 1933: 454), a Roman orator of the 1st century, who urges that ‘‘it is a good thing never to refute [one’s opponents] arguments in such a way that he seems a poor advocate’’. One way to do this is by concession which: expresses the fact that one gives a favourable receipt to one’s opponents real or presumed arguments. By restricting his claims, by giving up certain theses or arguments, a speaker can strengthen his position and make it easier to defend, while at the same time he exhibits his sense of fair play and his objectivity. (Quintilian, 1921–1933: 488, cited in Couper-Kuhlen and Thompson, 2000: 383).
1
Dostie (2009) defines ‘pragmaticalization’ as follows: ‘‘The term pragmaticalization was introduced by Erman and Kotsinas (1993) in a study of the marker you know in English and its functional equivalent in Swedish. The term refers to a process of linguistic change in which a full lexical item (noun, verb, adjective or adverb) or grammatical item (coordinator, subordinator, etc.) changes category and status and becomes a pragmatic item, that is, an item which is not fully integrated into the syntactic structure of the utterance and which has a textual or interpersonal meaning’’. There has been considerable (and ongoing) debate with respect to the relationship between grammaticalization and pragmaticalization and whether pragmaticalization involves an intervening grammatical stage (see Dostie, 2004: 21–40). My understanding of this is that grammaticalization refers to a lexical item which develops grammatical uses, whereas pragmaticalization refers to the process whereby a lexical/grammatical item develops uses which are conversational (related to discourse strategies) rather than propositional. Processes of grammaticalization and degrammaticalization may be concomitant with pragmaticalization – there is very often some form of recategorization as terms pragmaticalize, though morphology is less often affected than in grammaticalization, as it is generally understood.
84
Kate Beeching
The crux of the argument in this chapter is that concession is associated with the notion of ‘exhibiting one’s sense of fair play’. The act of (really or fictively) taking others’ views on board, thus protecting one’s own and one’s interlocutors’ self-image becomes associated metonymically with concessive or adversative expressions. This may lead to the use of such expressions purely as politeness markers, a type of hedge or mitigator. The resulting items do not fit readily into canonical word classes and are generally termed Discourse Markers (DMs) or pragmatic particles/expressions. The essentially dialogic and refutational nature of prolepsis is underlined by a number of authors (Forget, 1994, 2001; Vincent and Heisler, 1999; Couper-Kuhlen and Thompson, 2000). For Vincent and Heisler (1999: 29), prolepsis is an argumentative structure with four constituents : the thesis, the conceded statement, the rejection of the concession and the reaffirmation of the thesis. They follow Danon-Boileau and Morel’s (1992: 7) definition of concession as une manoeuvre rhe´torique double . . . [qui] permet, sans incohe´rence, de se faire l’avocat de l’une [des deux the`ses oppose´es] tout en apportant son assentiment a` l’un des arguments qui militent en faveur de l’autre a dual rhetorical strategy . . . which, without being incoherent, allows the speaker to support one (of two opposing arguments) whilst at the same time agreeing with one of the aspects militating in favour of the other.
Concession is used in two ways, according to Vincent and Heisler, either restrictively – to rectify a statement (thus falling into the domain of correction and reformulation) – or contrastively – to minimize an argument which runs counter to the thesis being proposed (thus falling into the domain of contrast and antithesis). Such definitions presuppose that concession is a logical, argumentative, process involving the weighing up of counter-arguments and their resolution in the third and fourth constituents (the rejection of the concession and the reaffirmation of the thesis). Vincent and Heisler (1999: 25) point out that, in the oral corpora from Quebec, it is rare for both the third or fourth constituents to be present, one or other is present and combines the two functions, the absent constituent being inferred. Of key interest is the fact that the ‘correct’ conclusion is sometimes left somewhat vague – the mechanism can thus be used to leave the door open to negotiation. Speakers can nuance the expression of their opinions, allowing for grey areas of indecision or for contradictory aspects of life to be expressed, but left unresolved. Couper-Kuhlen and Thompson (2000) illustrate the extent to which procatalepsis is part of our everyday conceptualization of events. They describe what they call the Cardinal Concessive which occurs in spontaneous exchanges, as follows (2000: 387): (5)
A: B Y
X That guy’s liver is not healthy Xu As much as he’s abused his liver (He’s abused his liver a lot) He’s still as healthy as an ox
Procatalepsis and the etymology of hedging and boosting particles
85
X and Y are understood by participants as potentially contrasting. The cardinal concessive is made up of three moves X – a statement; Xu – a concession and Y – a potential refutation. The concession in this case pivots on counter-expectation. Someone whose liver is not healthy might reasonably be expected to be unhealthy overall. B accepts that the ‘guy’ has abused his liver, but comes back with a refutation of the potential inference concerning his overall health, exclaiming that he is still healthy as an ox. In another example, Bee and Ava (on the phone) have been joking about each other’s knowledge of where the other is (Couper-Kuhlen and Thompson, 2000: 387) (6)
Bee: X Xu Bee: Y
I FINALLY said something right (0.2) YOU ARE HOME (hhh) Ava: yeah I BELIEVE so. PHYSICALLY anyway yea: h not MENTALLY though
Ava so strongly sets up a contrast between ‘physically and ‘not physically’ that Bee can collaboratively provide the Y of the Cardinal Concessive schema, in this case that she is not home mentally. Couper-Kuhlen and Thompson argue that it is the shared knowledge of, and familiarity with, the Cardinal Concessive which enables Bee and Ava to collaborate in the joint production of this pattern. The very fact that Bee can make the contrast explicit creates solidarity between the speakers. It is only in relatively recent works that the face-enhancing potential of concession in everyday interaction has been recognized, that aspect of procatalepsis which has been summed up in the words ‘fair play’. Forget (2001: 1160) argues that the crux of rhetorical or persuasive language is that it is interactional and works on the assumption that the other person in the engagement has rights and should be looked after. In this, it obeys general rules to do with politeness. She quotes Kerbrat-Orecchioni (1997: 15): ‘‘Polite communication consists above all in putting forward other people’s interest before one’s own’’. Forget suggests that one aspect of prolepsis is to activate competition between speakers, pitting one argument against another. There is, however, another interpretation: anticipating alternative points of view could be seen as reinforcing solidarity between speakers. Cette fois, le sce´nario e´voque´ reposerait sur la connivence entre les participants; le locuteur arrive a` deviner la position de son interlocuteur et a` satisfaire ses de´sirs avant qu’ils ne soient exprime´s – justement par l’activation d’un sche´ma culturel reconnaissable-, comme cela se pre´sente dans d’autres situations re´gle´es par les normes de la politesse (une personne cherche a` s’asseoir : on lui approche une chaise ou meˆme on lui ce`de sa place).
86
Kate Beeching
In this case, the scenario brought to mind is based on collusion between the participants; speakers manage to guess what their interlocutors’ point of view is and to satisfy their desires before they have been expressed – precisely through the activation of a recognizable cultural script – as happens in other situations regulated by norms of politeness (a person goes to sit down: someone brings up a chair or even gives up their own seat).
This paper argues that the metonymic contiguity of terms like though with contexts relating to the negotiation of meaning and the Cardinal Concessive frame so commonly invoked in everyday interaction has led them to be used in situations where X and Xu are unexpressed. Though then retains a concessive sense – but it functions simultaneously as a hedge and as a booster. What is more, this procataleptic tendency is a cross-linguistic phenomenon and one which affects similar adversative and concessive conjunctions and adverbs in different languages. In this chapter, I propose to illustrate this thesis with reference to the etymology of four pragmatic particles: French quand meˆme, English though, Scottish but and German aber. 3.1. Quand meˆme (‘all the same’) I first became interested in the way that an expression could be simultaneously hedging and boosting in my diachronic study of quand meˆme in French – Beeching (2005). The grammaticalization and pragmaticalization of quand meˆme has been a long drawn-out process, stretching over a number of centuries. Originally a fusion of the temporal conjunction quand (‘when’) with the reinforcer meˆme (‘even’), the coalesced form underwent semantic change as it grammaticalized, moving from a concessive conjunction to an adverbial. In the 17th century, quand mesme occurs exclusively as a coordinating conjunction along with the conditional tense and is concessive as in the following example: (7)
Je pre´pare un discours qui la pourroit toucher Quand mesme au lieu d’un coeur elle auroit un rocher. (Du Ryer, Pierre, Les vendanges de Suresne, 1636, page 62, Acte 1, sce`ne iv (vi)) ‘I’m preparing a speech which should move her Even though she had a rock in place of a heart.’ By the 20th century, quand meˆme is used almost exclusively as an adverb. It can retain an adversative thrust – but also has a relational usage. The adversative sense is used within a logical construction in three parts: P but quand meˆme Q. Moeschler and de Spengler (1981) give the following example: (8)
La rivie`re e´tait en crue mais le pont ne s’est pas quand meˆme e´croule´. ‘The river was in flood but all the same the bridge did not fall down.’
Procatalepsis and the etymology of hedging and boosting particles
87
A river in flood leads to the expectation that it might wash away the bridge; but introduces the contrasting phrase which refutes this expectation and ‘all the same’ recognizes and reinforces the concession to the implicit expectation that ‘all the same’ (despite the fact that we might expect the bridge to fall down) it did not. P but quand meˆme Q corresponds to Couper-Kuhlen and Thompson’s Cardinal Concessive: P but quand meˆme Q X
Xu
Y
Quand meˆme appears to encapsulate within itself a potential concession and thus simultaneously hedges and boosts Q. The first adverbial uses of quand meˆme begin to appear in the 19th century. Overall, in this century, in the theatrical works archived in FRANTEXT, as I charted in Beeching (2005), 54% of occurrences are conjunctions and 42% are adverbs. Most of these adverbs retain a strong adversative sense. If the evidence from the theatrical works contained in FRANTEXT can be relied upon to reflect contemporary spoken usage, at least to some extent, it is only in the early years of the 20th century that the adverb begins to pragmaticalize and to be used in ways which are not clearly adversative. In mid-20th and 21st century spontaneous spoken French, which we can investigate using spoken corpora such as the ESLO, the Beeching Corpus and the CRFP, quand meˆme frequently occurs WITHOUT the ‘P mais’ segment described by Moeschler and de Spengler. This usage is illustrated in example 9. (9)
2
Interviewer: oui . et dans la fac- on de parler franc- ais est-ce que vous trouvez qu’ il y a des diffe´rences entre les gens qui viennent de milieux diffe´rents? Interviewee: ah oui . oui oui ben oui oui . vous avez vous avez quand meˆme mettons un inge´nieur il aura quand meˆme pas le meˆme parler qu’ un ouvrier . il a davantage de diction davantage euh vous comprenez ? euh il a quand meˆme une une e´tude plus pousse´e que moi j’ ai e´te´ jusqu’ au certificat d’e´tudes . alors ( tandis) ( qu) ‘ ( eux) ils ont quand meˆme e´te´ euh remarquez que je cause peut-eˆtre moins peut-eˆtre pas tellement bien mais il y en a enfin qui causent encore plus mal que moi . oh oui . c- a de´pend le milieu ou` l’ on vit. (ESLO Corpus, Interview 105, utterances 465–4662)
To improve legibility, I have stripped out the tags which provide information about speaker identity and metalinguistic comments on pause length and so forth which are included in the original transcription.
88
Kate Beeching
‘Interviewer: yes . and in the way that people speak French do you find that there are differences between people who come from different backgrounds? Interviewee: ah yes . yes yes well yes yes . you’ve got you’ve got [quand meˆme] well shall we say an engineer he will have [quand meˆme] well not the same way of speaking as a worker . he has better diction euh you know ? euh he has [quand meˆme] well a better level of education than me I’ve just got GCSEs so (while them) they have [quand meˆme] well been euh mind you I may speak perhaps less perhaps not so well but there are well people who speak even worse than me . oh yes . it depends on the milieu you live in.’ 3 Some of the questions that form part of the sociolinguistic interview employed by the ESLO team in 1968–1971, such as this one, may appear somewhat surprising 40 years on. Here, the interviewer poses a direct question concerning social class differences in language. The interviewee’s reply includes four occurrences of quand meˆme. In none of these examples is it possible to retrieve the explicit X or Y terms of the Cardinal Concessive. In each case, the speaker seems to imply that there is a possible alternative point of view but that this is overwhelmed by counter-evidence. In mettons un inge´nieur il aura quand meˆme pas le meˆme parler qu’un ouvrier (‘shall we say an engineer he will have [quand meˆme] well not the same way of speaking as a worker’), the sense of quand meˆme might be paraphrased ‘whatever anyone might say to the contrary’. In the procataleptic framework, it could be expanded to: ‘I concede that there may be a view to the contrary of this but we have to admit that an engineer does not speak in the same way as a worker’. In Beeching (2005), I call this the ‘relational’ as opposed to the ‘adversative’ sense of quand meˆme. This relational sense is sensitively brought out by Grieve (1996), whose words seem to capture the manner in which quand meˆme can serve face needs in preempting an anticipated objection or criticism on the part of the interlocutor. In speech it is a tactical gambit which, by sketching an apparent attenuation of what might be sensed as the impropriety of an affirmation, can enable the reinforcement of the latter. (Grieve, 1996: 417).
In ‘euh il a quand meˆme une une e´tude plus pousse´e’ (‘euh he has [quand meˆme] well a better level of education than me’), quand meˆme seems to provide an apology or excuse – the speaker provides a rationale for the engineer’s ‘better diction’ which lies in his higher level of education.
3
DMs are notoriously difficult to translate – I have retained [quand meˆme] in the translation and rendered it using well, though this does not reproduce the full effect of quand meˆme in the context, which retains a hint of ‘all the same’.
Procatalepsis and the etymology of hedging and boosting particles
89
A paraphrase translation might be of course or he’s had a better education, though. Implicit in the interviewee’s words is the notion of opprobrium attached to not ‘talking proper’ – this implicit understanding is contradicted by pointing up the fact that the engineer is better educated, so the notion of blame is inappropriate. Quand meˆme provides a highly condensed manner of refuting an implicit understanding, of implying an implicit adversative. Grieve also notes this aspect of quand meˆme in saying: [Quand meˆme] facilitates what has been called la mise en acceptabilite´ d’une contradiction (Moeschler and de Spengler, 1981: 10). That is, it offers a justification for the statement it accompanies, even a sort of excuse or apology for it. But thereby it too has an adversative quality, faint and implicit, in that it hints at contradicting an assumed objection.
The final occurrence of quand meˆme in example 9 occurs at a point where the speaker breaks off mid-thought to nuance her remarks and avoid false modesty; she begins to repeat the fact that engineers have had a better education than she has, punctuating this statement with quand meˆme, but breaks off to say ‘remarquez . . . .’ (‘mind you’). After declaring that she may speak less well . . . , there is a syntactic break and the speaker reformulates and modulates the force of her utterance to say, ‘less, perhaps not so very well’. She then picks up her thread to add that there are, however, others who speak even worse than she does. Quand meˆme is at a stage of pragmatic ambiguity – it is certainly hard to pin down the exact pragmatic force of it, native speakers have some difficulty in saying what such markers mean and its pragmatic force seems to subtly change according to the context. It hedges and boosts at the same time, by suggesting an opposing point of view and contradicting it. Roulet et al. (1985: 82) consider such cases to be a type of implicit diaphony: l’e´nonciateur inte`gre et subordonne a` son discours pour le rejeter, un contre-argument qui n’a pas e´te´ ne´cessairement e´nonce´ par le destinataire, ou en tout cas qui n’apparaıˆt pas dans le co-texte. speakers integrate and subordinate to their argument, in order to reject it, a counter-argument which has not necessarily been uttered by their interlocutor, or in any case which does not appear in the co-text.
The use of the terms ‘argument’ and ‘counter-argument’, however, are perhaps misleading. They suggest that speakers trade in intellectual argumentation, pitting one thesis against another in the manner we might expect from Greek or Roman orators. As Forget (2001) points out, there is much in ordinary everyday language which is not ideational, or even textual, to use Halliday’s (1978) terminology, but is, rather, interpersonal. Conventions of social interaction which include taking the views of others into consideration, negotiating meaning and attending to face needs make the recruitment of language items for such purposes extremely likely – the terms undergo
90
Kate Beeching
grammaticalization and semantic bleaching and attain a stage where they are pragmatically ambiguous and ripe for pragmatic enrichment. Once such concessives or adversatives free themselves of both the strict logical P but Q structure and the strong concessive or adversative sense and serve to simultaneously concede and reinforce, they can be used in a far larger number of contexts to hedge and boost; in other words they become words used to mediate solidarity and social relations, rather than words employed in propositional contexts as logical connectors. My hypothesis is, then, that in everyday spontaneous conversation, adversative or concessive conjunctions or adverbs such as but, though, aber, doch, obwohl, pero, aunque, mais, pourtant and quand meˆme, bu`guo` and da`nshı` in Chinese and (da)kedo4 in Japanese can be used and often are used in a non-propositional, relational manner as pragmatic particles. In these cases, they are divorced from the canonical X, Xu, Y Cardinal Concessive pattern, but resume it by reproducing Xu in a highly elliptical manner and serve to simultaneously hedge and boost the co-text in the procataleptic manner described by Quintilian; the underlying ‘meaning’, that of showing a sense of ‘fair play’ becomes an integral part of the sense of the concessive item. 3.2. Though Though appears to follow a similar pattern to quand meˆme. It seems, however, that adverbial though (with an adversative sense) antedated the concessive conjunction (al)though and derived from Old Norse/o/which is related to Old High German/doh/and thence to Modern German doch. The adverbial form is syntactically constrained to clause-final position, though not always sentence- or utterance-final position. Like quand meˆme, though exploits the rhetoric of counterexpectation, as the core sense expresses the relation between two opposed facts or circumstances which co-occur, despite their apparent incongruity. Frequencies of occurrence in the British National Corpus suggest that though occurs more frequently in more recent, more spontaneous and more personal texts than in more formal speech and writing. Figures 5.1–5.3 reveal that the highest frequency of though in the BNC is in email correspondence, followed by personal letters and spontaneous conversation. The count of occurrences of though in the BNC did not in this case differentiate conjunctival from adverbial though or canonical adversative from relational usages. It might perhaps be reasonable to assume, however, that higher frequences correlate with bleached, relational usages and not with propositional, adversative usages, whose lexical strength constrains flexibility of implementation. While it might be possible to test this hypothesis by conducting a quantitative
4 Anecdotal evidence gained in personal communication at the IPRA conference, Gothenburg, 8–13 July 2007 suggests that both Chinese bu`guo` (however) and Japanese (da)kedo (but, however) may be used as pragmatic particles. On dakedo, see also Ondera (1995).
Procatalepsis and the etymology of hedging and boosting particles
REGISTER
91
SPOKEN
FICTION
NEWS
ACADEMIC
NONFIC MISC OTHER MISC
TOKENS
4,286
7,034
2,725
5,400
5,291
8,603
SIZE (MW)
10.33
16.19
10.64
15.43
16.63
28.39
PER MIL
414.7
434.3
256.2
350.0
318.1
303.0
Figure 5.1. Frequency of occurrence of though in the BNC, all registers
OTHER MISC
ADM ADV BIO COM EML ES1 ES2 HAN INS INL LET1 LET2 MSC POP REL
TOKENS
0
5
2,224 2,879 456
SIZE (MW)
0.01 0.22 0.56 3.53
12
3.76 0.21 0.15 0.06 1.16 0.55 0.44 0.05
0.07
9.15
7.38 1.12
PER MIL
0
239
76
243
390
55
45
81
1,574 897
446
202
948
79
539
8
144
35
30
82
150
65
149
40
762
407
LIST OF NAMES FOR [MISC] REGISTERS ADM ADMIN
COM
COMMERCE
ES2
ESSAY_UNIV
INL
INSTRUCTIONAL
ADS
ADVERT
EML
EMAIL
HAN
HANSARD
LET1
LETTER_PERSONAL POP
POP_LORE
BIO
BIOGRAPHY ES1
INSTITUT_DOC
LET2
LETTER_PROF
REL
RELIGION
ESSAY_SCHOOL INS
MSC
MISC
Figure 5.2. Frequency of occurrence of though in the BNC, OTHER MISCELLANEOUS registers analysis of, for example, all 4286 occurrences of though in the spoken part of the BNC, each example would require a detailed qualitative analysis, relying on a substantial section of co-text to divide all adversative usages (where both P and Q elements are present) from all relational usages (where either P or Q is present but not both). In practice, due to the very nature of such usages, in which P or Q may be implicit or the usage may be pragmatically ambiguous, the researcher has few formal criteria to draw on and finds herself relying on intuition and interpretation. The possibility of reliably allocating each occurrence neatly to an ‘adversative’ or ‘relational’ subcategory is thus excluded.
92
Kate Beeching
SPOKEN
DIS
DOC
NEW
CLS
CNS
CNV
CRT
DEM
INT
IOH
COM
HUM
TOKENS
185
8
94
88
24
3,004
13
3
23
122
3
9
SIZE (MW)
0.76
0.04
0.26
0.43
0.14
4.21
0.13
0.03
0.12
0.82
0.02
0.05
PER MIL
244
193
360
205
174
714
102
94
186
150
199
177
SPOKEN
NAT
POL
SOC
MTG
PRL
DEB
SRM
SP+
SP-
SPO
TUT
UNC
TOKENS
3
4
27
343
20
30
19
14
95
22
55
78
SIZE (MW)
0.02
0.05
0.16
1.38
0.10
0.28
0.08
0.20 0.46
0.03
0.14
0.42
PER MIL
132
79
169
249
208
106
231
70
660
384
185
204
LIST OF NAMES FOR [SPOKEN] REGISTERS DIS
BRDCST_DISCUSSN CNV
DOC
BRDCST_DOCUM
CRT
CONV
COM LECT_COMMERCE
COURTROOM
HUM LECT_HUMANITIES
NEW BRDCST_NEWS
DEM DEMONSTRATN
NAT
LECT_NAT_SCIENCE
CLS
CLASSROOM
INT
INTERVIEW
POL
LECT_POLITICAL
CNS
CONSULT
IOH
INTRVW_ORAL_HIST SOC
LECT_SOC_SCIENCE
MTG
MEETING
SP-
SPEECH_UNSCRIPTED
PRL
PARLIAMENT
SPO
SPORTSLIVE
DEB
PUB_DEBATE
TUT
TUTORIAL
SRM
SERMON
UNC
UNCLASSIFIED
SP+
SPEECH_SCRIPTED
Figure 5.3. Frequency of occurrence of though in the BNC, spoken registers
Procatalepsis and the etymology of hedging and boosting particles
SECTION
SPOKEN
FICTION
NEWSPAPER
ACADEMIC
MISC
1.4 15.3 21
9.0 44.6 401
93
SEE ALL SECTIONS
PER MIL SIZE (MW) FREQ
110.1 10.0 1097
44.4 15.9 707
11.4 10.5 119
Figure 5.4. Frequency of sentence-final though in the BNC. One way of approaching this is to search for though . (sic) in the corpus, using the full stop at the end to bring up all the examples of sentence-final though (as, although we know that not all relational thoughs are sentence-final, and not all sentence-final thoughs are relational, this does at least exclude conjunctival though). Figure 5.4 shows the distributional frequencies of these forms and it is perhaps unsurprising that sentence-final though is far more common in spoken than in other genres, and that it does appear fairly strongly in fiction, too. The following examples highlight some of the characteristics of though in the BNC. The collection of emails in the BNC is devoted almost exclusively to exchanges concerning football matches. (10)
Looking at games individually the performances have been good, and results good (‘‘cept QPR). However 3 draws means 6 points dropped. Being unbeaten is something positive though considering the last 4 or 5 games.
In example 10 (I must beg the reader’s indulgence here for any misinterpretations which arise from my ignorance of football terminology), it is clearly the case that though accompanies a P but Q structure in the canonical manner: results have been good, 6 points have dropped but the team is unbeaten. However, because though occurs at the end of the utterance – being beaten is something positive, though, the word appears to reinforce or boost the proposition whilst at the same time recognizing a restriction. (11)
with your ticket application, I don’t know if you can use them over the phone. Anyway I’ll try to sort this out and let everyone know. They look good value for 12 quid though. I’m suprised (sic) at the lack of SCUM bashing on the server. Shall I send a letter to the square
In example 11, the term to which the proposition is opposed is absent. In They look good value for 12 quid though, though is used in a very similar way to quand meˆme to raise the spectre of a
94
Kate Beeching
possible objection in the mind of the hearer, to hedge and simultaneously boost the accompanying utterance. This usage is yet more evident in the occurrences drawn from the spontaneous conversation files in the BNC. (12)
Yeah, try a pickled onion Here half a one int it? Mm, the last Mm, good that vinegar though cos I sliced some onion, I like it like that, do you?
In example 12, the P but Q logical structure is absent. There is, however, an implicit counterargument concerning the quality of the vinegar – though suggests that it is unexpected that the vinegar is good. In the context, though mediates self-deprecation: the speaker has clearly made the pickled onions himself/herself and does not wish to boast. S/he suggests, by using though that s/he might have expected the onions not to be perfect and thus respects the face-need to minimize one’s expertise (see Coates, 1997: 120 on ‘expertism’). As Kerbrat-Orecchioni (1997: 15) says, ‘‘it may be recommended in certain circumstances to deprecate [one’s faces] (to damage one’s own territory, or to belittle oneself)’’. The speaker is negotiating agreement with the hearer. This is reinforced by the tag question in ‘I like it like that, do you?’ Relational though is often accompanied by other pragmatic hedging and boosting expressions such as a bit, really or tag questions, as in examples 13–22: (13) (14) (15) (16) (17) (18) (19) (20) (21) (22)
Yeah. It’s a bit embarrassing though. Nice house though isn’t it? He was a devil though. That was a good shot though, wasn’t it? They played well though, didn’t they Ron? I was lucky though. it’s right though isn’t it? I mean why should you have to put up with it? it’s not very personal to give somebody a voucher though is it? What amazes me though he just don’t seemed to of. you know. I tell you what though.
In all of these examples, though can be omitted without loss to the propositional content of the utterance and appears to function in a similar way to the most bleached form of quand meˆme. This indicates that, in the spoken British English of the BNC, non-conjunctival though is very frequently relational. The examples also illustrate the fact that not all relational thoughs are sentence-final and we must therefore interpret the findings based on searches for though . with caution. Further evidence for the increase in the frequency of though usage can be adduced by searching for the term in the American Time Magazine corpus. Figure 5.5 shows that occurrences of though increased dramatically from the 1920s to the 1970s but that frequency has dropped back slightly in more recent years.
Procatalepsis and the etymology of hedging and boosting particles
SECTION
1920s
1930s
1940s
1950s
1960s
1970s
1980s
1990s
2000s
254.4 7.6 1942
295.6 12.7 3741
272.9 15.5 4217
379.3 16.8 6368
532.4 16.1 8561
558.9 13.6 7597
499.9 11.4 5685
496.0 9.7 4829
415.0 6.4 2667
95
SEE ALL SECTIONS
PER MIL SIZE (MW) FREQ
Figure 5.5. Though in the Time Magazine Corpus
SECTION
1920s
1930s
1940s
1950s
1960s
1970s
1980s
1990s
2000s
0.1 7.6 1
0.9 12.7 12
1.2 15.5 18
1.4 16.8 23
2.5 16.1 41
4.6 13.6 62
6.5 11.4 74
9.9 9.7 96
10.1 6.4 65
SEE ALL SECTIONS
PER MIL SIZE (MW) FREQ
Figure 5.6. Sentence-final though in the Time Magazine Corpus. A more focussed picture emerges when a search is made for though. (sentence-final though followed by a period). Figure 5.6 shows that, though frequencies are low with only 392 occurrences in the entire corpus, progression is very clear, with an increase in usage of 100% from the 1920s to the 2000s. An examination of the usages of the 65 examples of though . in the 2000s section of Time Magazine did not support the argument, made above, that though is rising in frequency due to its relational usages in this register. All of the examples, albeit often taken from direct speech, are adversative rather than relational. Examples include: (23) (24) (25)
Their men’s socks are amazing too – kind of expensive, though. But in TIME’s tests, the Discovery still got stuck on the edges of a rug and missed dust in corners. It’s fun to watch, though. Contemporary women no longer find hard-drinking men attractive. This too will change, though.
The evidence presented here suggests that sentence-final though is increasing in frequency across different varieties of English (both British and American English). It retains a contrastive sense in more formal written registers, where it remains relatively rare (10.1 occurrences per million words).
96
Kate Beeching
In informal and spoken British English registers, where occurrences are much higher (110 occurrences per million words), relational usages are more common. Though remains polysemous and can be used as a concessive conjunction. Where it is used adverbially and/or sentence-finally, it is concessive or pragmatically ambiguous, being both a recognition that there is an alternative point of view and underscoring that point of view, hedging and boosting. Because conjunctival though coexists with adverbial though, there is persistence of its core concessive meaning. Syntactically, its increasingly prevalent positioning on the right-periphery as a concessive adverbial appears to act as a bridge to its usage as a hedging/boosting relational device in everyday conversation. An alternative interpretation might be that the increasing frequency of though as a relational device in the spoken language has led to an increase in sentence-final adversative uses in the written language – the use of though imparts a demotic, conversational, tone. 3.3. Refutation and adversatives Unlike though, quand meˆme can constitute an utterance on its own, as a type of mitigated refutation, with a meaning similar to ‘Surely not!’ or ‘Really!’, exploiting once again the rhetoric of counter-expectation, as we see in the invented example (26). (26) A: B: A: B:
Jean-Pierre a re´ussi a` ses examens. Quand meˆme. Pete has passed his exams. *Though./Really.
Scholars working on but in English and its equivalents in other languages (Marconi and Bertinetto, 1984; Blakemore, 2002: 92ff on Italian ma; Anscombre and Ducrot, 1977; Nyan, 1999; Carel, 2002 on mais; Lo¨tscher, 1989; Diewald and Fischer, 1998 on aber) have tended to focus on their logical, propositional force, considering, for example, the lack of differentiation between the senses of aber and sondern which are subsumed under but in English or taking a Gricean or relevance-theoretic approach to implicature. Nyan (1999: 215–220) discusses the interesting case of Mais mange! (‘But eat!’) discussed in Ducrot et al. (1980: 128), in which there is an implicit anterior context of a child being enjoined to eat but neglecting to do so, to which ‘mais mange’ (‘but eat’) is opposed. She also raises the question of a usage informally described as the ‘mais d’invitation’, illustrated in ‘Mais venez donc dıˆner!’ (‘But come to dinner then!’). In this case, there is no anterior context to which mais is opposed and Nyan suggests (1999: 220) that, though the evocation of an anterior event to which this is opposed cannot account for the degree of politeness which would not figure without mais, cet effet de sens se comprend sans difficulte´ si l’on interpre`te ‘mais’ comme pre´sentant le fait que D. doit venir dıˆner comme e´vident.
Procatalepsis and the etymology of hedging and boosting particles
97
this sense is easily understood if one interprets but as presenting the fact that D. should come to dinner as self-evident.
Mais (but) seems to me to be at an early stage of pragmaticalization (not least on the formal grounds that it is syntactically constrained) and I will not consider it in great detail here. However, the example Nyan discusses is, to my mind, more easily interpretable as a hedging/boosting politeness particle than as a marker of mutual manifestness. Such an interpretation sits well in the politeness-and-procataleptic framework described above. In the following, I will focus on modal uses of but and aber. 3.4. But But derives from the West Germanic compound *be, bi (by) þ *utana (from without, outside), in other words, from a sense which is clearly oppositional, using a space metaphor to describe the intellectual concept of opposition (in vs. out). Generally considered in contemporary speech to be an adversative conjunction, coordinating two contradictory facts or circumstances, but is employed in a post-rhematic adverbial manner in some regional registers of English (including Australian English5). Beeching (2007: 108) gives 22 examples of clause-final but collected in Dumbarton in SW Scotland (this somewhat unusual use of but is also a feature in the North of England, particularly Tyneside6). Like quand meˆme and though, but, even when it is utterance-final, is lexically polysemous and can be pragmatically ambiguous. It retains its canonical adversative sense in some contexts but not in others. In order to show that they are indeed polysemous, it has to be demonstrated that the P term in the P but Q construction is not in evidence. For this, a reasonably large section of co-text is required. In the 22 examples of utterance-final but in the Dumbarton list, some are clearly oppositional but some are clearly relational. I have selected three below to illustrate this. (27)
A: Real Madrid are playing tonight. B: They’re not televising it but. In example 23, there are clearly two terms an X and a Y term with an intervening underlying assumption that, if a football match is being played, it will be televised. B refutes the implicature and but serves as an adversative refutation marker: but could clearly be replaced by however here – they are not televising it, however.
5
My thanks go to the member of the audience who pointed this out at my presentation on this topic to the UK-Cognitive Linguistics Association, Cardiff, August 27–30, 2007. 6 See the Newcastle Electronic Corpus of Tyneside English (NECTE).
98
Kate Beeching
(28)
A: I’ll ring Jim and let him know what’s happened. B: I don’t think we’re going to need him. A: I’ll ring him and let him know but.
As I argued in Beeching (2007: 93), the use of but in example 28 is more ambiguous. A makes a proposal, B gives a covert refutation of A’s proposal – the proposition I don’t think we’re going to need him contains the underlying performative ‘Don’t phone him ¼ there’s no need, we won’t need him’ – A on the other hand nuances B’s response – I’ll ring him and let him know but – this but could be replaced logically by nonetheless or all the same or however but in the context but softens A’s rebuttal of B’s rejection of his initial proposal. It recognizes B’s objection but reinforces the fact that A will go ahead regardless. This but occupies the middle ground between the propositional and relational sense of but. example 29 illustrates the fully pragmaticalized form of but (29)
I was teaching the wee man to ride his bike the day. It’s good to see him cycling but.
There is no obvious contrast between I was teaching the wee man to ride his bike the day and it’s good to see him cycling. But at the end of the utterance could be described as interjective in the same way as utterance initial but in Aye but it’s grand. But intensifies or reinforces the accompanying expression. 3.5. Aber German aber is canonically a coordinating conjunction corresponding to but in English: (30)
Ich mo¨chte ins Ausland fahren ‘I would like to abroad to travel’ aber ich habe kein Geld. ‘but I have no money.’ I’d like to go abroad but I have no money.
It also, however, has a modal usage: (31)
Das ist aber scho¨n! ‘That is but beautiful!’ Isn’t that beautiful!
In her (2005) doctoral thesis on translations of Scott Fitzgerald’s The Beautiful and the Damned, Marion Winters explores the extent to which German modal particles, such as aber are used in dialogue. She claims on page 68 that: The basic semantic meaning of aber is that of surprise, which tends to be a rather positive surprise (Bublitz, 1978; Weinrich, 1993). Aber refers to an expectation (Bublitz, 1977; Ehlich, 1984)
Procatalepsis and the etymology of hedging and boosting particles
99
which is exceeded (Lu¨tten, 1977). Thus, it shows the adversative feature that is characteristic of its homonym aber as a conjunction, and is therefore labelled by Weinrich with change (Wendung). Aber has an intensifying effect and often occurs in exclamations. It also commonly occurs in requests/demands or – followed by auch – in reproaches. Aber is very strongly related to dialogue.
The examples she gives clearly demonstrate the simultaneously hedging and boosting quality of aber used as a modal particle, which echoes the use of rhetorical question forms such as aren’t you?, tag questions, such as don’t you? and interjections such as Well and Why in Scott Fitzgerald’s English. (32)
(33)
‘‘Das ist aber lieb von dir, dab du ‘‘That is but sweet of you, that you dir Sorgen um mich macht.’’ you DAT worries about me make.’’ (HCOl.709) FSF: ‘‘Why, aren’t you sweet to worry about me.’’ Du hast aber Mumm!’’ Rief sie ‘‘You have but guts!’’ shouted she bewundernd aus. admiringly out. (HC01.283) FSF: ‘‘Well, you are spunky!’’ she exclaimed admiringly.
Aber exploits the rhetoric of counter-expectation in such a way as to reinforce the surrounding co-text. By including aber, the speaker suggests that she might not have expected her interlocutor to be worried about her in example 32 or to have been courageous in example 33, hence the concept of ‘surprise’. (34)
‘‘Jetzt ho¨r mir aber mal zu’’, ‘‘Now hear me DAT but DM to’’, plapperte Anthony mit schwankender Stimme . . . chattered Anthony with wavering voice (HCO2.1513) FSF: ‘‘Now, listen, ‘‘chattered Anthony unsteadily . . . In example 34, used with the imperative, aber also suggests counter-expectation, but this time, the expectation is that the listener might not listen – aber therefore contains an element of surprise – this may not be what you are expecting – and serves to paradoxically hedge and boost the imperative. (35)
‘‘Du trinkst aber auch sta¨ndig’’ ‘‘You drink but DM constantly’’ stellte sie unvermittelt fest.
100
Kate Beeching
put she suddenly forth. (ROGa.705) FSF: ‘‘You drink all the time, don’t you?’’ she said suddenly. Finally, in example 35, aber accompanies a reproach. Once again, its force derives from its assertion of a counter-expectation, expressing surprise at the situation and thus reinforcing – or potentially hedging – the implied criticism. The use of aber parallels the way that tag questions like don’t you in English can have a number of functions, adding a softening or challenging tone (Holmes, 1995: 80–82), hedging or reinforcing. Though the pragmatic force of modalizing items such as aber and don’t you? are not easily pinned down, may be ambiguous and may not be translationally equivalent, the fact that aber has been selected to render boosting or hedging expressions in English is revealing. The ‘surprise’ notion inherent in modal uses of aber appears to capture the concessive ‘although I would not have expected it’ and reinforces or boosts the argument; coupled with an imperative or a reproach, on the contrary, the effect is a hedging and simultaneously boosting one in the manner of the Cardinal Concessive or procataleptic frame. By conventionally suggesting that there might be a counter-expectation (using a tag question, or including aber), the speaker taps in to a recognized frame of reference, the procataleptic described by Quintilian, restricting one’s claims, strengthening one’s position while maintaining a sense of ‘fair play’. The hedging and boosting qualities of quand meˆme, clause-final but, though and aber have a common etymology, deriving as they do from conjunctions which introduce an adversative or concessive element. The question as to why clause-final but exists in some varieties and not in others is difficult to answer and it is similarly difficult to ascertain why mais is less fully pragmaticalized than aber. One reason may be related to word order constraints in French vis-a`vis German. Conjunctival aber may move around, as in example 36: (36)
deshalb daraus kann dann aber because-of-that out of it can then but wieder etwas entstehen again something come out ‘but something can then come out of it again because of that’ (Datenbank gesprochenes Deutsch, Grundstrukturen: Freiburger Korpus, F23)
Even in its contrastive sense, aber can be placed, not clause-initially but later in the syntagmatic chain. This may ease the transition from a more propositional (syntactically constrained) to a more pragmatic (syntactically unconstrained) usage. French word order constrains contrastive conjunctival mais to a clause-initial position. There would, however, be no reason why mais should not be clause-final or post-rhematic, as there is a syntactic and rhythmic slot which begs to be filled here in French (cf. quoi, see Andrews, 1989). This has, however, to my knowledge, not occurred in any variety of French.
Procatalepsis and the etymology of hedging and boosting particles
101
4. CONCLUSION Investigating the nature of the pragmaticalization process which leads from M1 to M2 is a ticklish process, as the semantic change may be lengthy and remain for centuries at the M1/M2 stage characterized by polysemy and pragmatic ambiguity. It is generally recognized that change occurs gradually, step by step, in spoken interaction, and may spread unevenly in different styles or registers of the language (see Hansen, 2005: 67, on enfin; Heinz, 2006, on prosodic features) – and there may be geographical variation in the extent to which terms pragmaticalize (witness Dostie, 2009, on coudon, pis, alors and (c- a) fait que in varieties of Canadian and hexagonal French; and the existence of clause-final but in Australia, in Tyneside and the Dunbartonshire area of Scotland but not elsewhere). A fruitful avenue for further enquiry is suggested by Vincent and Heisler (1999: 20), who note that concession in Quebec French is often introduced by OK. They remark in a footnote: Nous conside´rons que OK est grammaticalise´ comme connecteur concessif a` cause de sa fre´quence d’usage (voir Heisler, 1996). Il serait inte´ressant de ve´rifier si ce processus de grammaticalisation, de marqueur d’accord vers la concession et la prolepse, est re´pandu dans d’autres langues. We consider that OK is grammaticalized as a concessive connector because of its frequency of usage (see Heisler, 1996). It would be interesting to check whether this process of grammaticalization, from agreement marker towards concession and prolepsis, is prevalent in other languages.
Beeching (2009) argues that bon (‘good’) is used increasingly as a mitigating expression in contemporary hexagonal spoken French, a position which is also held by Bre´mond (2004), and, for Chinese hao (‘good’), by Wang and Tsai (2005). Though some authors (Hansen, 1998: 254; Wang and Tsai, 2005) relate this mainly to discourse coherence, a primary motivation for the shift (from agreement to concession) may be related to face needs and the maintenance of the semblance of agreement: (Pre-)closing signals such as hao may be regarded as a sub-variety of mitigating expressions used in conversation, that is, the desire to agree or appear to agree with the addressee, which leads to mechanisms for pretending to agree (Brown and Levinson, 1987: 113). Such expressions serve the twofold function of keeping the conversation going in a systematic manner while allowing the conversationalists to preserve either the reality or the appearance of cooperation (Wang and Tsai, 2005: 231).
In their study of bien (‘good’, ‘well’) in Spanish and French, Waltereit and Detges (2007) argue that different diachronic pathways result from the use of a term in different pragmatic strategies:
102
Kate Beeching
DMs derive from strategies related to the joint coordination of interaction, while modal particles derive from strategies which refer to the status of a given proposition to the ongoing discourse. According to their (2007: 68) definitions of DMs and modal particles, members of the quand meˆme family (if we can call it that) start as DMs, yet their original utterance-initial (adversative) function has given way to an intersubjective modalizing one because of their interactional usages. Traugott’s (1982) prediction that textual (discourse-marking) usages precede intersubjective or expressive ones seems to have greater explanatory power in this case. The Cardinal Concessive, as we have seen, allows for the joint coordination of interaction while considerations of politeness confirm the hedging/boosting (modal) usage. The gradual entrenchment of the procataleptic usage of quand meˆme, though, but and aber described and illustrated in this chapter goes one step further along the route to pragmaticalization than the maintenance of cooperation in interaction which one sees in the use of OK, well, bon or bueno. Quand meˆme, though, but and aber have developed hedging/boosting pragmatic usages from similar concessive or adversative semantic ‘core’ meanings; their polysemies differ, their etymologies differ and their position in the clause differs, but the fundamental cognitive reflex depending on the Cardinal Concessive, which is part of our everyday conceptualization of events, is very similar. I would argue, too, like Forget (2001), that the underlying motivation for the recruitment of such usages is universal, arising from social interactional exigencies related to questions of face.
ACKNOWLEDGMENT I acknowledge the contribution made by the editors and anonymous reviewers, whose insightful comments led me to clarify the arguments presented and to add further data analysis. I am of course responsible for any remaining errors or misinterpretations.
REFERENCES Andrews, B. (1989). ‘‘Terminating devices in spoken French’’. International Review of Applied Linguistics in Language Teaching XXVII (3): 193–216. Anscombre, C. and O. Ducrot (1977). ‘‘Deux mais en franc- ais’’. Lingua 43: 23–40. Beeching, K. (2005). ‘‘Politeness-induced semantic change: The case of ‘‘quand meˆme’’’’. Language Variation and Change 17 (2): 155–180. Beeching, K. (2007). ‘‘A politeness-theoretic approach to pragmatico-semantic change’’. Journal of Historical Pragmatics 8 (1): 69–108. Beeching, K. (2009). ‘‘Sociolinguistic factors and the pragmaticalization of bon in contemporary spoken French’’, in K. Beeching, N. Armstrong and F. Gadet (eds.), Sociolinguistic variation in contemporary French. Amsterdam: John Benjamins, 215–229.
Procatalepsis and the etymology of hedging and boosting particles
103
Blakemore, D. (2002). Relevance and linguistic meaning. The semantics and pragmatics of discourse markers. Cambridge: Cambridge University Press. Bre´mond, C. (2004). ‘‘La petite marque bon, l’indice d’un accord en cours de ne´gociation’’. Travaux de Linguistique 48: 7–19. Brown, P. and S. Levinson ([1978] 1987). Politeness. Some universals in language usage. Cambridge: Cambridge University Press. Bublitz, W. (1977). ‘‘Deutsch aber als Konjunktion und als Modalpartikel’’. Linguistische Arbeiten 50: 199–209. Bublitz, W. (1978). Ausdrucksweisen der Sprechereinstellung im Deutschen und Englischen. Tu¨bingen: Max Niemeyer Verlag. Carel, M. (2002). ‘‘Occupe-toi d’Ame´lie. Emploi contrastif de mais et illustration’’. Cahiers de Linguistique Franc- aise 24: 160–205. Coates, J. (1997). ‘‘Womens’ friendships, women’s talk’’, in R. Wodak (ed.), Gender and discourse. London: Sage, 245–262. Couper-Kuhlen, E. and S. A. Thompson (2000). ‘‘Concessive patterns in conversation’’, in E. Couper-Kuhlen and B. Kortmann (eds.), Cause-condition-concession-contrast. Cognitive and discourse perspectives. Berlin: Mouton de Gruyter, 381–410. Danon-Boileau, L. and M.-A. Morel (1992). ‘‘Operations e´nonciatives et valeurs argumentatives: e´tude de trois marqueurs concessifs’’, in Approches e´nonciatives de l’e´nonce´ complexe. Louvain: Peeters, 7–19. Diewald, G. and K. Fischer (1998). ‘‘Zur diskusiven und modalen Funktion der Partikeln aber, auch, doch und ja in Instruktionsdialogen’’. Linguistica Ljubljana 38 (1): 75–99. Dostie, G. (2004). Pragmaticalisation et marqueurs discursifs. Analyse se´mantique et traitement lexicographique. Bruxelles: De Boeck & Larcier. Dostie, G. (2009). ‘‘Discourse markers and regional variation in French. A lexico-semantic approach’’, in K. Beeching, N. Armstrong and F. Gadet (eds.), Sociolinguistic variation in contemporary French. Amsterdam: John Benjamins, 201–214. Ducrot, O., D. Bourcier and S. Bruxelles. (1980). ‘‘Mais occupe-toi d’Ame´lie’’, in Les mots du discours. Paris: Editions de Minuit, 93–130. Ehlich, K. (1984). ‘‘Eichendorffs aber’’, in W. van Peer and J. Renkema (eds.), Pragmatics and stylistics. Leuven: Acco, 145–192. Erman, B. and U.-B. Kotsinas (1993). ‘‘Pragmaticalization: The case of ba’ and you know’’. Studier i Modern Spra˚kvetenskap 10: 76–93. Forget, D. (1994). ‘‘Anticipation et argumentation: la prolepse’’. Revue Que´be´coise de Linguistique 23 (1): 61–77. Forget, D. (2001). ‘‘Figures, politesse et organisation textuelle’’. Journal of Pragmatics 33: 1157–1172. Grieve, J. (1996). Dictionary of contemporary French connectors. London: Routledge. Halliday, M. A. K. (1978). Language as social semiotic: The social interpretation of language and meaning. London: Edward Arnold. Hansen, M.-B. M. (1998). ‘‘The semantic status of discourse markers’’. Lingua 104: 235–260. Hansen, M.-B. M. (2005). ‘‘From prepositional phrase to hesitation marker: The semantic and pragmatic evolution of French enfin’’. Journal of Historical Pragmatics 6 (1): 37–68. Hansen, M.-B. M. (2008). Particles at the semantics/pragmatics interface: Synchronic and diachronic issues. A study with special reference to the French phasal adverbs. Oxford: Elsevier. Heinz, M. (2006). ‘‘La prosodia come marcatura discorsiva: osservazioni sulla differenziazione prosodica di due tipi di testo in italiano’’, in M. Drescher and B. Frank-Job (eds.), Les marqueurs discursifs dans les langues romanes. Frankfurt am Main: Peter Lang.
104
Kate Beeching
Heisler, T. (1996). Une analyse fonctionnelle et sociolinguistique de la particule OK en franc- ais montre´alais, Me´moire de maıˆtrise. Que´bec: Universite´ de Laval. Holmes, J. (1995). Women, men and politeness. London: Longman. Kempson, R. (2001). ‘‘Pragmatics, language and communication’’, in M. Arnoff and J. Rees-Miller (eds.), The handbook of linguistics. Oxford: Blackwell, 394–427. Kerbrat-Orecchioni, C. (1997). ‘‘A multi-level approach in the study of talk-in-interaction’’. Journal of Pragmatics 7 (1): 1–20. Lo¨tscher, A. (1989). ‘‘Implikaturen und Textfunktionen im Gebrauch von Konnektiven des Gegensatzes’’. Linguistische Berichte 121: 215–239. Lu¨tten, J. (1977). Untersuchungen zur Leistung der Partikeln in der gesprochenen deutschen Sprache (Go¨ppinger Arbeiten zur Germanistik 218). Go¨ppingen: Verlag Alfred Ku¨mmerle. Marconi, D. and P. M. Bertinetto (1984). ‘‘Analisi di ) ma *. (Parte prima : semantica e pragmatica)’’. Lingua e Stile 19 (2-3): 223–258. Moeschler, J. and N. de Spengler (1981). ‘‘Quand meˆme: de la concession a` la re´futation’’. Cahiers de Linguistique Franc- aise 2: 93–112. Nyan, T. (1999). ‘‘Vers un sche´ma de la diffe´rence: le cas de mais’’. Journal of French Language Studies 9: 211–238. Ondera, N. (1995). ‘‘Diachronic analysis of Japanese discourse markers’’, in A. Jucker (ed.), Historical pragmatics. Pragmatic developments in the history of English. Amsterdam: Benjamins, 393–437. Quintilian (1921–1933). The instituto oratoria of Quintilian. Translated by H. E. Butlet. New York: G. P. Putnam’s Sons. Roulet, E., A. Auchlin, J. Moeschler, C. Rubattel and M. Schelling (1985). L’articulation du discours en franc- ais contemporain. Bern: Peter Lang. Sweetser, E. (1990). From etymology to pragmatics. Metaphorical and cultural aspects of semantic structure. Cambridge: Cambridge University Press. Traugott, E. C. (1982). ‘‘From propositional to textual and expressive meanings: Some semantic-pragmatic aspects of grammaticalization’’, in W. P. Lehmann and Y. Malkiel (eds.), Perspectives on historical linguistics (Amsterdam studies in the theory and history of linguistic science IV. Current issues in linguistic theory, Vol. 24). Amsterdam: Benjamins, 245–272. Traugott, E. C. and R. Dasher (2002). Regularity in semantic change. Cambridge: Cambridge University Press. Vincent, D. and T. Heisler (1999). ‘‘L’anticipation d’objections: prolepse, concession et re´futation dans la langue spontane´e’’. Revue Que´be´coise de Linguistique 27 (1): 15–31. Waltereit, R. and U. Detges (2007). ‘‘Different function, different histories. Modal particles and discourse markers from a diachronic point of view’’. Catalan Journal of Linguistics 6: 61–80. Wang, Y.-F. and P.-H. Tsai (2005). ‘‘Hao in spoken Chinese discourse: Relevance and coherence’’. Language Sciences 27: 215–243. Weinrich, H. (1993). Textgrammatik der deutschen Sprache. Mannheim: Dudenverlag. Winters, M. (2005). ‘‘A corpus-based study of translator style: Oeser’s and Orth-Guttmann’s translations of F. Scott-Fitzgerald’s The Beautiful and the Damned’’. Unpublished PhD thesis. Dublin City University, Dublin.
CORPORA Beeching Corpus of Spoken French. http://www.uwe.ac.uk/hlss/llas/iclru/corpus.pdf British National Corpus (BNC). http://CORPUS.BYU.EDU/BNC
Procatalepsis and the etymology of hedging and boosting particles
Corpus de Re´fe´rence du Franc- ais Parle´ (CRFP). http://www.up.univ-nrs.fr/delic/crfp Datenbank gesprochenes Deutsch. http://dsav-wiss.ids-mannheim.de/ E´tude Sociolinguistique d’Orle´ans (ESLO). http://bach.arts.kuleuven.ac.be/lancom FRANTEXT. http://atilf.atilf.fr/ Newcastle Corpus of Tyneside English (NECTE). http://www.ncl.ac.uk/necte/ TIME Corpus of American English. http://corpus.byu.edu/time/
105
Current Trends in Diachronic Semantics and Pragmatics Edited by Maj-Britt Mosegaard Hansen and Jacqueline Visconti r 2009 by Emerald Group Publishing Limited. All rights reserved.
6 CENTRAL/PERIPHERAL FUNCTIONS OF ALLORA ‘OVERALL PRAGMATIC CONFIGURATION’: A DIACHRONIC PERSPECTIVE
AND
Carla Bazzanella and Johanna Miecznikowski
1. INTRODUCTION This paper deals with Italian allora (possible English translations are ‘at that time’, ‘then’, or ‘so’),1 a polysemic and polyfunctional connective and discourse marker (DM) that is highly frequent in Present-day Italian. Previous corpus-based research (Bosco and Bazzanella, 2005) has contrasted uses of allora in Old Italian with recent uses in spoken and written Italian. A comparison of the Padua corpus, which consists of texts written in Florentine from 1250 to 1300 (cf. Renzi, 2002), and CORIS, documenting Present-day written Italian, has shown that in Present-day Italian the original distal temporal meaning of allora (‘at that time’, ‘then’) is much less frequent than in Old Italian. In the Padua corpus, 96% of the tokens of allora have been found by Bosco and Bazzanella (2005) to have temporal meaning. In CORIS, this proportion decreases to 52%, whereas different other
1
Cf. lat. illa hora. Existing etymological dictionaries (Battaglia, 1961: 335; Cortellazzo and Zolli, 1979: 41) simply suggest that Italian allora derives from Latin ad illam horam ‘at that time’. However, since this phrase is not attested either in classical or in medieval Latin texts, it seems better to assume that the Latin temporal ablative illa¯ hora¯, which gives rise to French lors, has survived in Italian only within expressive prepositional phrases – most prominently in allora, but also in other forms such as Old Lombard inlor(a), allo (see von Wartburg, 1952, vol. 4: 478). The same phenomenon is attested in French alors, from a` lors, as well as in Old Fr. adonc ‘alors, puis’, from a` donc. See, e.g., Rey (1992: 2055) s.v. lors. (A. Garcea, personal communication, December 2007).
108
Carla Bazzanella and Johanna Miecznikowski
semantic values and functions increase in number, e.g., the modal use in hypothetical if-then constructions (‘then’). Moreover, new values and functions are attested, e.g., the use as an inferential marker (‘so’) or as an interaction structuring marker, close to English markers such as ‘well’, ‘now’, ‘ok’. In these new uses, allora has lost its referential functions as a connective expressing temporal or consequential relationships between states of affairs, and has acquired new cognitive, textual and interaction-related functions as a connective expressing relations between utterances and speech acts. In virtue of this loss of functions on the level of propositional content, the more recent uses of allora are to be considered as DM. The different DM uses of allora in contemporary Italian have been further investigated by Bazzanella et al. (2007, 2008a, b) and Miecznikowski et al. (2009),2 focused on spoken rather than written language. In this paper, we will examine in more detail the diachronic changes that have occurred, on the basis of an essentially qualitative, and for some periods only exploratory study taking into account several corpora of written and spoken Italian (see Section 2). To describe diachronic change, we will consider the lexical item under analysis as a prototypically organized category, whose center is defined by frequent types of use and abstract features that constitute their invariants, whereas less frequent uses and less clearly linguistically coded features associated with them are placed at the category’s periphery. In this view, diachronic change has to do with various types of uses and features shifting from the periphery to the center and vice versa, and with new uses entering into the category’s periphery (for an application of prototype theory to diachronic analysis, cf. also Geeraerts, 1997; Pons Borderı´a, 1998). In its early phases, the diachronic development of allora corresponds to a well-described grammaticalization path from (spatial)/temporal to what in Bosco and Bazzanella (2005) has broadly been termed ‘modal’, in accordance with other existing works in grammaticalization research.3,4 For the sake of this paper, we will strictly distinguish between modal meanings, which contribute to the propositional content of the utterance – in the case of allora, its use as a connective introducing the consequent of a conditional construction – and DM uses. We will specifically highlight those innovations which, within temporal and modal uses, have prepared the recruitment of allora into the Italian DM lexicon (Sections 3 and 4). We will trace back some central cognitive, textual, and interactional functions of contemporary allora to the marker’s preDM stages, when they were only peripherically present. We will try to identify both semantic changes and, in Section 4, context-related phenomena that are likely to have contributed to the strengthening of these functions.
2
Cf. Hansen (1997) for the similar case of French alors. Cf., inter alia, Traugott (1982), Sweetser (1990), Brinton (1996), Hansen (1998, 2008), Traugott and Dasher (2002), Bazzanella et al. (2005), Hopper and Traugott (2003), and Fischer (2006). 4 Further Italian examples are anzi (Bazzanella, 2003), gia` (cf. Bazzanella et al., 2005), addirittura, perfino, qualora (Visconti, 2006) or, outside the domain of DMs, Lombardi Vallauri (2010). 3
Central/peripheral functions of allora and ‘overall pragmatic configuration’
2. ALLORA
FROM
OLD ITALIAN
TO
109
PRESENT-DAY ITALIAN: AN OVERVIEW
2.1. Old Italian For Old Italian, the research has been mainly based on the Padua corpus (cf. Section 1 above). Moreover, we have taken into account OVI (Opera del vocabolario italiano, http:// www.csovi.fi.cnr.it/), which includes 1780 texts from the origins up to 1375, looking for early examples of allora used with inferential and textual functions. At this stage, allora is characterized by relative syntactic freedom. Apart from its use in conditional constructions, which we will consider in more detail in Section 3, it has mainly temporal meaning. A typical example is example 1, where allora is a temporal adverb that sets the scene for a state of affairs in the past with imperfective aspectual properties: (1)
Usavansi allora le medaglie, in Firenze, che le due valevano uno danaio piccolo. ‘In those days the Florentines used the medaglia as currency, two of which were worth a penny.’ (Novellino, novel 96, cf. Padua corpus)
Another common variant is connective allora introducing an event with perfective aspect (cf. example 2). This variant expresses temporal consecution of events in the past or future. Interestingly, in contrast to other connectives like poi it underlines also the existence of a causal relationship between the consecutive events, especially the specific relationship between actions and reactions. In example 2, this relationship is foregrounded thanks to the gerund udendo questo. (2)
‘‘ . . . insegnatemi campare, che´ potete, et io saro` vostro marito, e terrovi onorevolmente’’. Allora la donna, udendo questo, innamoro` di questo cavaliere e disse: ‘‘Io faro` cio` che voi mi comanderete [ . . . ]’’. ‘‘Teach me the way to survive; I know you can, and I will become your husband and hold you in the highest esteem.’’ Then the woman, hearing this, fell in love with this knight and said, ‘‘I will do what you ask [ . . . ]’’ (Novellino, novel 59, cf. Padua corpus)
As a connective, Old Italian allora can have text-structuring functions. It contributes for instance to signaling the change of perspective from one agent to the next in narrative contexts, or is used to introduce the point of a story or a new narrative sequence. In the overall picture, however, these text-structuring functions are peripheral. In particular, there seems to be no use of allora in Old Italian in which textual functions dominate clearly the marker’s referential functions in the temporal or modal domain.
110
Carla Bazzanella and Johanna Miecznikowski
2.2. Allora from the 15th to the 19th century For the period from 1400 up to the 19th century, we have conducted a very limited exploratory study, taking into account the following literary texts: Giovanni Boccaccio, Decameron (1349– 1351); Pietro Bembo, Prose della volgar lingua (1525); Baldesar Castiglione, Il cortegiano (1528); Giovanni della Casa, Il Galateo (1558); Carlo Goldoni, a selection of 10 plays (18th century); Manzoni, I promessi sposi (1840). All texts are available online (http:// www.classicitaliani.it/) and contain dialogues including direct speech, in which occasionally one finds DMs. With regard to allora, its uses as a DM seem to arise only in the 18th/19th century. We have found the first examples in direct speech or free indirect speech in Manzoni, where some tokens of allora introduce a specific speech act. Consider example 3, where allora introduces a piece of advice: (3)
‘‘Cosa m’importa a me che uomo sia o non sia un altro, quando quel pover’uomo che non c’e` piu`, era quello che sapeva le nostre cose, e aveva preparato tutto per aiutarci?’’ – ‘‘Allora, bisogna aver pazienza.’’ ‘‘What does it matter to me what a man is or is not, when that good man, who’s no longer here, was he who knew all our affairs, and had made preparations to help us?’’ – ‘‘Then you must have patience.’’ (Promessi sposi, cap. XVIII; translation: Manzoni, 1997–2008)
In this type of use, which is still common in contemporary Italian, allora establishes a consequential relation between a speech act by the speaker and the preceding co-text uttered by an interlocutor. It may be paraphrased as ‘‘under these circumstances (or: given what you just said) I utter speech act X’’. 2.3. Present-Day Italian For Present-Day Italian, we refer to several corpora of spoken Italian5 and to a novel by Giuseppe Culicchia, Paso Doble, which is stylistically close to informal spoken language and contains a large amount of direct speech.
5 A selection of transcripts from the following corpora has been used: CLIPS and API (cf. www.clips.unina.it); C-ORAL-ROM (cf. Cresti and Moneglia, 2005); LIP (cf. De Mauro et al., 1993); TUT (cf. http://www.di.unito.it/Btutreeb/); AVIP (cf. Bertinetto, 2001); CHILDES (cf. Bortolini and Pizzuto, 1997); IPAR (cf. Albano Leoni and Giordano, 2005); one edition of the talk-show ‘‘Amici di sera’’ (12-3-1997) (cf. Bazzanella et al., 2008b).
Central/peripheral functions of allora and ‘overall pragmatic configuration’
111
Contemporary allora has expanded its functional spectrum. It has developed uses that do not or only minimally contribute to propositional content, but mainly perform functions
in the domain of cognitive and argumentative relations, more precisely, inferential functions as a marker of a conclusion based on premises given in the preceding context;
and in the interactional domain, in particular turn-introduction, turn eliciting, or as a textstructuring Gliederungssignal (Gu¨lich, 1970).
Consider example 4, taken from an interrogation. PRE, the policeman, uses utterance-final allora to signal that its host utterance is an inference based on what CUS, the accused, has just said: (4)
CUS [ . . . ] questa e` una questione / che non c’entra onienteW PRE oe adesso quiW c’e` un altro mandato / allora? CUS certo // CUS [ . . . ] that’s a question / that has nothing to do owith itW PRE oand now hereW there’s another mandate / you mean? CUS that’s right // (C-Oral-Rom/inatla_02)6
Example 5 is taken from a medical presentation. The talk is given by ROS, while GEN intervenes occasionally as a chairperson. (5)
ROS
[turn several minutes long, in which ROS refers to several clinical studies and explains a typology of stenoses] ovviamente / la stenosi grave / resta grave // GEN rimane sotto / oa un centimetro quadratoW ROS ocentimetro quadratoW // avanti // allora / la [inbreath] (il) successivo evento / risale a tre anni dopo [continues on this topic for several minutes] ROS of course / the severe stenosis/is still severe // GEN it is below / oone square centimeterW ROS osquare centimeterW // next please // ok / the [inbreath] (the) next event / has occurred three years later (C-Oral-Rom/inatco_3)
In this case, allora signals the beginning of a new segment of discourse, distinct from the preceding one in several regards: as to discourse genre (from a medical state of the art to a specific anamnesis), as to visual support (a new slide), and, strictly locally, as to
C-Oral-Rom transcription conventions: o W ¼ overlap; / ¼ boundary of an intonation unit; // ¼ prosodically determined utterance boundary; [ ] ¼ comment.
6
112
Carla Bazzanella and Johanna Miecznikowski
participation framework (from an interaction with the chairperson to a monological part of the talk). Fragment 6, finally, is an example of turn-eliciting allora, exploited to intensify the expression of the speaker’s impatience: (6)
‘‘Vado a prenderglielo subito’’. Cercai e cercai e cercai. [ . . . ] ‘‘Allora, c’e` o non c’e` questo corso?’’ ‘‘I’ll go and fetch it.’’ I searched and searched and searched. [ . . . ]‘‘do you finally have that course or don’t you have it?’’ (Giuseppe Culicchia, Paso doble, p. 34)
3. THE DIACHRONIC DEVELOPMENT
OF ALLORA’S
USES
AS A
DM
The range of allora’s uses as a DM in Present-day Italian has been preceded by a series of innovations in diachronic development. These innovations, to which we already alluded briefly in the previous section, can be described as a gradual strengthening of certain features, changing the central or peripheral status of allora’s various uses and making new uses emerge. We will now look at three features in more detail.
3.1. Increase of connectivity and changes of scope The first diachronic change we would like to underline is the strengthening of connectivity, which in the case of allora involves (a) an increasing preference for anaphoric scope over states of affairs mentioned in an immediately preceding discourse segment, and (b) a loss of syntactic freedom, with an increasing tendency toward utterance-initial position. These developments have given rise to an iconic configuration in which allora is placed between two segments, denoting two entities that are construed as being consecutive – be it in the domain of temporal reference, conditional relations, inference, or textual and interactional sequentiality. This evolution may be illustrated by the case of constructions with a subordinate clause introduced by quando (i.e., ‘when’ or ‘whenever’) and a main clause containing allora. In this construction, allora expresses simultaneity, posteriority, or a conditional relation. In Old Italian, the syntactic position of allora in the main clause is relatively free, and the main clause is often placed before the subordinate clause. A good example is example 7, which contains one clause final and one clause initial occurrence of allora, both relating a state of affairs with generic aspect to another generic state of affairs that precedes it temporally and is a
Central/peripheral functions of allora and ‘overall pragmatic configuration’
113
necessary condition for it. The quando-clause expressing the condition is postponed in both cases:7 (7)
Et Cassiodero disse: che l’apparecchiamento e` buono allora quando egli e` fatto con lunghi pensieri; percio` che tutte le cose subite, per cierto, son dubbiose, e allora vede l’uomo, ch’egli a` mal fatto quando egli sente ‘l danno che gli de’ venire. ‘And Cassiodero said: that the preparation [of a battle] is good + when it results from long reflection; because, for sure, all hasty things are dubious, and [only] then does man see that he has done wrong, when he gets aware of the damage he risks to suffer.’ (Andrea da Grosseto, Trattati morali di Albertano da Brescia volgarizzati, 1268, cf. OVI)
Our exploratory study suggests that this type of structure has become a more peripherical use of allora already by the 16th century – even if partly lexicalized immediately adjacent allora quando (cf. the first token in example 7) seems to have resisted at least until the 18th century. In contemporary Italian, all variants with allora referring cataphorically to a postponed quandoclause have virtually disappeared. The increase in (iconic) connectivity and the corresponding marginalization of cataphoric or long-distance anaphoric uses has probably influenced the development of the contemporary DM uses, because it makes it easy for hearers to recover the entities in the marker’s scope from co- and context, reinforcing contextual enrichment. Contextual enrichment in turn favors changes in scope, from components of propositional content (time periods and states of affairs) to cognitive, textual, and actional components of the on-going communicative event. It is not by chance that in contemporary spoken Italian utterance-initial allora is most polyfunctional and frequently fulfils structuring functions, taking scope over text and interaction sequences (cf. example 5), whereas utterance-final allora (e.g., in example 4) tends to have narrower scope and has a more restricted functional range. 3.2. From consecution to the expression of a consequent construed as inferable/expectable A second interesting diachronic shift is related to allora’s use in hypothetical contexts. In these contexts, over time, allora came to introduce consequents that are easily inferable on the basis of previous discourse, laying the ground for the subsequent uses of the lexeme as an inferential DM. In Old Italian, allora frequently expresses a relation of consecution between a necessary condition and its consequent. In these cases, it means ‘only then’, often both in a modal and in a temporal sense (‘not yet now, only after the condition has been fulfilled’ – cf. example 7). In other
7 The construction shares important characteristics with relative subordination, in particular with the lexicalized variant allora che ‘(at the time) when’.
114
Carla Bazzanella and Johanna Miecznikowski
cases, however, this meaning component is backgrounded and allora is used to reinforce the contrast between one hypothetical scenario (cf. Dancygier and Sweetser, 2005) and an alternative scenario mentioned before. This kind of use is illustrated by example 8, where the contrast is made explicit by ma (‘but’): (8)
[ . . . ] se ll’ uditore non fosse al tutto turbato contra noi, ben potemo acquistare benivoglienza per principio. Ma s’ ei troppo malamente fosse straniato ver noi, allora ne conviene rifuggire a ‘‘insinuation’’. ‘[ . . . ] if the hearer is not completely upset with us, we can obtain his benevolence by a direct introduction. But if he is too hostile against us, then we have to resort to ‘‘insinuation’’ [ . . . ].’ (Brunetto Latini, Rettorica, ca. 1260/61, cf. OVI)
The two uses of allora in conditional constructions differ as to the way in which the consequent is related to the preceding co-text. The (temporal)hypothetical variant signaling the consequence of a necessary condition is typically integrated into a co-textual configuration in which the consequent is not easily inferable on the basis of the preceding context. The contrastive type of allora, on the other hand, often introduces conclusions of complex argumentations, which are easily inferable on the basis of the preceding context. As example (8) illustrates, their inferability is due to an argumentative schema of the type ‘if p then q, but if :p allora r’, in which an evident contrast between the two scenarios limits the range of possible r’s to those that are :q. Our restricted data suggest that, around the 16th century, this use comes to prevail over the noncontrastive one. It is plausible to assume that the recurrent association of modal allora with inferability has promoted the integration of this feature at the periphery of the marker’s functional potential. Some evidence compatible with this hypothesis is provided by Goldoni’s plays (18th century), in which the argumentative value of modal allora is sometimes rhetorically exploited in (conditional) promises or permissions and promise/permission-eliciting questions. Allora then underlines the intersubjectively demonstrated reasonableness of the conditional action plan and the speaker’s/interlocutor’s commitment to it. Consider the following example: (9)
T: G: T: G:
T: G:
Lo vuo` ricever solo. Ve l’ho da dir cantando? Voi mi mandate via. Sı` signore, vi mando. So che scherzate, amico, percio` non me n’offendo: Dovete restar solo, e` ver, non lo contendo. Ma quando il forestiere sia stato un pezzo qui, Potro` venire allora? Signor no. Signor sı`. (parte.)
Central/peripheral functions of allora and ‘overall pragmatic configuration’
115
T: G: T: G:
‘I want to receive him alone. Do I have to sing to convince you?’ ‘You are sending me away.’ ‘Yes, sir, I am.’ ‘I know you are joking, my friend, therefore I do not offend myself. You [2nd pers. pln.] must be left on your own, it is true, I do not contest it. But after the foreigner being here a while Will I/I will be allowed to come, then?’ T: ‘No, sir.’ G: ‘Yes, sir. (leaves)’ Goldoni, Torquato Tasso (1755, second act, scene 8a) In this example, G construes a contrast (cf. the contrastive marker ma) between the present situation (a), in which he accepts T’s request of being left alone (1–5), and a future situation (b), for which he hopes to obtain T’s permission to be present, in virtue of the fact that T’s request will have been fulfilled at that time (6–7). As in other contrastive uses of allora in conditional constructions, the consequent introduced by allora is easily inferable. Moreover, situation (b) is a desirable option for G and the entire conditional construction may therefore be interpreted as an argumentation in favor of a standpoint (‘you want to be alone with the foreigner; so once this desire will be fulfilled you should allow me to join you’). In contexts such as the one illustrated by example 9, the inferability of the consequent implies its plausibility as a standpoint and is a particularly salient feature of the utterance, whereas the speculative functions of conditional constructions (weighing of different alternatives) becomes less important. Such uses are therefore likely to have favored allora’s reinterpretation as a marker of plausible conclusions. In Manzoni, finally, we find instances of elliptic and emphatic uses of allora, like in example 10: (10)
Alle volte, una cagione momentanea puo` fare un’impressione che par che deva durar sempre; e quando poi la cagione cessa, e l’animo si muta, allora . . . ‘There are times when a passing cause may make an impression that seems at the moment sure to be lasting; but afterwards, when the cause is removed, and the mind calmed, then . . . ’ (I promessi sposi, cap. X; translation: Manzoni, 1997–2008)
The persuasive effect of these uses, which appeal to the hearer to complete the conditional construction, depends directly on allora’s projecting a highly inferable consequent. The above-described semantic-pragmatic shift within hypothetical constructions is crucial for understanding the rise of allora’s uses as a (non hypothetical, cf. Section 3.3) DM. Not only is the postulation of inferability and expectability a central feature of the uses introducing speech acts (cf. example 3) and inferences (cf. example 4). Expectability on co-textual grounds may be considered relevant also for more recent text and interaction structuring allora, by which speakers
116
Carla Bazzanella and Johanna Miecznikowski
punctuate discourse and interaction in accordance with intersubjectively shared action scripts and genre conventions (cf. example 5), differing in this respect from other structuring markers; or for variants of allora that elicit next moves by the interlocutor construing them as legitimately expectable on the basis of such shared scripts (example 6 can be analyzed in this way).
3.3. Deictic-modal shift and evidential properties A third, and very significant, semantic change is the recent possibility of using allora with temporally proximal, factual reference. In Old Italian, allora signals temporal and/or modal distance, i.e., it refers to states of affairs that are situated in the past, in the future, in a world construed as possible/counterfactual, or that have generic or distributive aspect. These uses correspond to English ‘then’ ‘at that time’. They exclude reference to the speech point (in Reichenbach’s, 1947 terms) and in some cases establish even a clear contrast to it – e.g., in example 1, where allora ‘in those days’ is opposed to the here and now ideally shared by the narrator and the reader. In contrast, the DM uses of allora typical of Present-day Italian refer to the speech point and to the on-going interaction, without evoking either temporally distal or non-factual states of affairs. Allora then corresponds to English ‘now’/’at this time’/’under these conditions’. This deictic-modal shift is certainly related to connectivity and to scope changes, as sketched in Section 3.1 above. Moreover, specific developments concerning the use in hypothetical contexts must have played a role. One possible promoting force of allora’s semantic shift toward the origo is an increasing use of conditional allora with reference to specific events in the near future (e.g., for promises, cf. Section 3.2 above). In this section, we would like to focus on a different phenomenon, though, i.e., on reference to others’ discourse. The spoken data we have analyzed show that when using allora to introduce inferences or speech acts, speakers do not refer to premises given in their own discourse (cf. example 4), and if they do, the effect is polyphonic distantiation. This evidential property is present already in the first speech act introducing instances of allora we found, i.e., in Manzoni (19th century), where allora refers back to the interlocutor’s discourse (cf. example 3); it might have arisen out of dialogically coconstrued conditional reasoning as illustrated, e.g., by Goldoni’s plays (cf. example 9). We hypothesize that in the early DM uses, modal distance (non factuality) and polyphonic distance (reference to Other’s discourse) are co-present features (‘‘if it is as you/X say(s), I perform speech act X’’). Later, probably, the feature ‘Other’s discourse’ gradually came to dominate over the feature ‘non factuality’ (‘‘given what you say, I perform speech act X/I infer X’’). As a result, the utterance X introduced by allora has acquired temporally proximal, factual reference.
Central/peripheral functions of allora and ‘overall pragmatic configuration’
117
This likely semantic change in inferential and speech act introducing variants of allora, which is not uncommon in conditional constructions in general,8 can be seen as having prepared the diachronically posterior development of the marker’s contemporary structuring uses. These have completely lost any modal meaning and may be seen as manifesting a further reanalysis of polyphonic functions in terms of textual and interactional functions, by introducing a segment (e.g., a conversational move, a phase of interaction, or a new topic) considered as a relevant, and often script motivated (cf. Section 3.2), continuation of the preceding sequential development.
4. CONTEXTS OF USE, RELEVANT PARAMETERS, CONFIGURATION’
AND
‘PRAGMATIC
In the preceding sections, we have looked at meaning shifts that have contributed to the development of allora’s modern uses as a DM. Such meaning shifts, however, are not autonomous diachronic processes, but are to be seen as interrelated with changes as to contexts of use. If it has been recognized in grammaticalization research that semantic innovations often spread from quite specific co-texts to a wider range of possible co-texts, the pragmatic characteristics of such specific inference-generating co-texts are not always included in the focus of the analysis. In our view, they should be, in a fruitful interchange between semantic and pragmatic analysis. In this last section, we will therefore identify some contextual parameters that are likely to have played a role in the diachronic development of allora – neglecting, for reasons of space, other relevant aspects such as prosody (cf. Bazzanella et al., 2008a). A first relevant parameter is genre or activity type.9 The prototypical temporal use of allora in Old Italian is tightly related to narration. In contrast, it is essential to recognize that the hypothetical uses, whose key role we have stressed repeatedly, typically occur in dominantly argumentative genres or activity types. The strengthening of the features ‘high inferability/ expectability’, in particular, cannot be understood without taking into account the textual and pragmatic features of such genres, especially – as mentioned in Section 3.2 – the presence of complex argumentations leading to an explicit conclusion, and the relevance of conditional reasoning for making decisions on future actions. A second set of relevant parameters is connected with oral mode and face-to-face interaction. Certain changes seem indeed closely linked to the cognitive needs of spoken language: e.g., the increasing tendency toward utterance-initial iconic placement of connective allora, which may be 8
Cf., e.g., Bazzanella (1989) for ‘‘correlative’’ uses of Italian se (‘if’) and Miecznikowski (2008) with regard to evidential and attenuating uses of the conditional form in Italian and French. 9 Cf. Bazzanella et al. (2008b) for a corpus-based investigation of the relationship between activity types and various uses DM uses of allora in contemporary Italian.
118
Carla Bazzanella and Johanna Miecznikowski
seen as a kind of simplification; or the more recent development of the potential to project new segments of discourse, related to recipient-designed planning in on-going speech production. Another important property of face-to-face interaction is co-construction. Very generally, the continuous need for coordination and for the construction of intersubjectivity in face-to-face interaction makes the spatio-temporal deictic center particularly important for speakers (cf., among many others, Lyons, 1977), an attentional focus reflected by allora’s (modally mediated) deictic shift from ‘then’ to ‘now’. More specifically, the functional shift observed in conditional constructions – from a de re connective expressing a relation between states of affairs to a de dicto connective signaling an act of relating utterances (via action motivation or inference) – has crucially to do with that act becoming an interactional issue. Evidence for this claim may be found in the 18th/19th century use of conditional allora in questions (in Goldoni and Manzoni) and in elliptical statements (cf. example 10). It is indeed by requesting the hearer to complete his/her reasoning that the speaker – presupposing the hearer’s ability to do so – foregrounds and reinforces the feature ‘inferability/expectability’, already potentially present in conditional allora.10 Co-construction in face-to-face situations may therefore be seen as a driving force of semantic and pragmatic change in the pre-DM stage, whereas only later it has become a linguistically coded functional domain of the DM allora, first on the level of content (especially by expressing cohesion combined with polyphonic distantiation from the interlocutor’s discourse) and then on the level of activity coordination.
5. CONCLUSION The case of allora shows quite clearly the relevance of the interface between semantics and pragmatics (cf. Turner, 1999; Hansen and Turner, 2006), investigated here in relation to diachronic change, and the necessity to integrate semantic and pragmatic analysis. In our view, important aspects of the tight relationship between semantics and pragmatics can be captured through the notion of ‘overall pragmatic configuration’: Depending on the specific configuration selected within a certain context (e.g., type of text, role of the speaker, etc.), some functions may be primarily involved in determining the DM interpretation, while other functions may be involved only secondarily. In other words, all the functions co-exist, to a certain extent, but depending on the configuration that is activated, some will more powerfully participate in determining the DM interpretation. (Bazzanella et al., 2007: 26).
10 It is worth noting that questions, elliptic statements, and the combination of both – interrogative allora occupying an entire turn – are still frequent structural contexts of allora as a DM in contemporary Italian.
Central/peripheral functions of allora and ‘overall pragmatic configuration’
119
In Section 3, we have identified a series of ‘‘emergent structures’’ (Hopper, 1987; Bybee et al., 2001) in discourse, i.e., uses of allora in which the co- and context activates features which eventually become associated with the marker itself. We have argued that both the frequency of occurrence of such structures and the particular pragmatic salience of the features in question play a role in their becoming part of the marker’s functional potential. In Section 4, we have insisted on the importance of contextual parameters in this process. The notion of overall pragmatic configuration turns useful in this respect. It underlines that ‘‘emergent structures’’ of this kind correspond to complex configurations of a large range of cotextual and contextual parameters together with the propositional content of the utterance and other constitutive dimensions of the pragmatic perspective (e.g., speaker’s attitude, interaction, and multimodality, cf. Bazzanella, 2006, 2008). These parameters synergically interact to construe utterance meaning and constrain the contribution of any single linguistic item to it, favoring both the selective activation of certain functional properties and inferences that may lead to a reanalysis of the item’s functional potential. This helps us understand why not only co-text, but also certain aspects of context can become relevant driving forces of semantic change.
REFERENCES Albano Leoni, F. and R. Giordano (eds.) (2005). Italiano parlato. Analisi di un dialogo. Napoli: Liguori. Battaglia, S. (1961). Grande Dizionario della lingua italiana. Torino: UTET. Bazzanella, C. (1989). ‘‘Il se correlativo nell’italiano scritto contemporaneo’’. Lingua e stile XXIV (1): 33–55. Bazzanella, C. (2003). ‘‘Dal latino ante all’italiano anzi: la ‘deriva modale’’, in A. Garcea (ed.), Colloquia Absentium. Studi sulla comunicazione epistolare in Cicerone. Torino: Rosenberg, 123–140. Bazzanella, C. (2006). ‘‘DMs in Italian: Towards a ‘compositional’ meaning’’, in K. Fischer (ed.), Approaches to discourse particles. Amsterdam: Elsevier, 449–464. Bazzanella, C. (2008). Linguistica e pragmatica del linguaggio. Un’introduzione. Roma: Laterza. Bazzanella, C., C. Bosco, E. Calaresu, A. Garcea, P. Guil and A. Radulescu (2005). ‘‘Dal latino iam agli esiti nelle lingue romanze: verso una configurazione pragmatica complessiva’’. Cuadernos de filologia italiana 12: 49–82. Bazzanella, C., C. Bosco, A. Garcea, B. Gili Fivela, J. Miecznikowski and F. Tini Brunozzi (2007). ‘‘Italian allora, French alors: Functions, convergences, and divergences’’. Catalan Journal of Linguistics (special issue ed. by M. J. Cuenca) 6: 9–30. Bazzanella, C., C. Bosco, B. Gili Fivela, J. Miecznikowski and F. Tini Brunozzi (2008). ‘‘Polifunzionalita` dei segnali discorsivi, sviluppo conversazionale e ruolo dei tratti fonetici e fonologici’’, in M. Pettorino, A. Vallone, M. Vallone and R. Savy (eds.), La comunicazione parlata. Napoli: Liguori ed., 934–963. Bazzanella, C., C. Bosco, B. Gili Fivela, J. Miecznikowski and F. Tini Brunozzi (2008). ‘‘Segnali discorsivi e tipi di testo’’, in C. Bosisio, B. Cambiagli, E. Piemontese and F. Santulli (eds.), Aspetti linguistici della comunicazione pubblica e istituzionale. Perugia: Guerra, 239–265. Bertinetto, P. M. (ed.) (2001). AVIP – Archivio di varieta` di italiano parlato, 4 CD-Rom. Pisa: Ufficio Pubblicazioni della Classe di Lettere della Scuola Normale Superiore.
120
Carla Bazzanella and Johanna Miecznikowski
Bosco, C. and C. Bazzanella (2005). ‘‘Corpus linguistics and the modal shift in Old and Present-Day Italian: Temporal pragmatic markers and the case of ‘allora’’’, in C. D. Pusch and W. Raible (eds.), Corpora and historical linguistics. Tu¨bingen: Gunter Narr, 443–453. Bortolini, U. and E. Pizzuto (eds.) (1997). Il Progetto CHILDES-ITALIA-Contributi di ricerca sulla lingua italiana. Pisa: Edizioni del Cerro. Brinton, L. J. (1996). Pragmatic markers in English: Grammaticalization and discourse functions. Berlin: Mouton de Gruyter. Bybee, J., et al. (ed.) (2001). Frequency and the emergence of linguistic structure. Amsterdam: John Benjamins. Cortellazzo, M. and P. Zolli (1979). Dizionario etimologico della lingua italiana. Bologna: Zanichelli. Cresti, E. and M. Moneglia (eds.) (2005). C-ORAL-ROM. Integrated reference corpora for spoken romance languages. Amsterdam: John Benjamins. Dancygier, B. and E. Sweetser (2005). Mental spaces in grammar. Conditional constructions. New York, NY: Cambridge University Press. De Mauro, T., F. Mancini, J. Vedovelli and M. Voghera (1993). Lessico di frequenza dell’italiano parlato. Milano: Etaslibri. Fischer, K. (ed.) (2006). Approaches to discourse particles. Amsterdam: Elsevier. Geeraerts, D. (1997). Diacronic prototype semantics: A contribution to historical lexicology. Oxford: Clarendon Press. Gu¨lich, E. (1970). Makrosyntax der Gliederungssignale im gesprochenen Franzo¨sisch. Mu¨nchen: W. Fink. Hansen, M.-B. M. (1997). ‘‘‘Alors’ and ‘donc’ in spoken French: a reanalysis’’. Journal of Pragmatics 28: 153–187. Hansen, M.-B. M. (1998). The functions of discourse particles. A study with special reference to spoken French. Amsterdam: John Benjamins. Hansen, M.-B. M. (2008). ‘‘The semantics of particles: Advantages of a contrastive and panchronic approach. A study of the polysemy of French ‘de´ja`’ and Italian ‘gia`’’’. Linguistics 46 (3): 471–505. Hansen, M.-B. M. and K. Turner (eds.) (2006). Explorations in the semantics/pragmatics interface. Acta Linguistica Hafniensia, Vol. 38. Copenhagen, CA: Reitzels Forlag. Hopper, P. J. (1987). ‘‘Emergent grammar’’. Berkeley Linguistics Society 13: 139–157. Hopper, P. J. and E. C. Traugott (2003). Grammaticalization. 2nd rev. Edition. Cambridge: Cambridge University Press. Lombardi Vallauri, E. (2010), ‘‘Free conditionals in discourse: the forming of a construction. Linguisticae Investigationes 33 (1), forthcoming. Lyons, J. (1977). Semantics. Cambridge: Cambridge University Press. Manzoni, A. (1997–2008). I Promessi Sposi Or The Betrothed. Great Literature Online. Available at: http:// manzoni.classicauthors.net/IPromessiSposiOrTheBetrothed/(15 Jan, 2008). Miecznikowski, J. (2008). ‘‘Gli usi del condizionale nel parlato italiano e francese’’, in M. Pettorino, A. Vallone, M. Vallone and R. Savy (eds.), La comunicazione parlata. Napoli: Liguori ed., 865–902. Miecznikowski, J., B. Gili Fivela, and C. Bazzanella (2009). ‘‘Words in context. Agreeing and disagreeing with allora’’, in G. Gobber et al. (eds.), Word meaning in argumentative dialogue ( =L’analisi linguistica e letteraria, Vol. 16, No. 1), Special Issue, 205–218. Pons Borderı´a, S. (1998). Conexion y conectores. Valencia: Universitat de Valencia. Reichenbach, H. (1947). Elements of symbolic logic. London: Macmillan. Renzi, L. (2002). ‘‘Il progetto Italant e la grammatica del corpus’’. Verbum 4 (2): 271–294. Rey, A. (1992). Dictionnaire historique de la langue franc- aise, E´d. Dictionnaires Le Robert, Paris.
Central/peripheral functions of allora and ‘overall pragmatic configuration’
121
Sweetser, E. (1990). From etymology to pragmatics. Metaphorical and cultural aspects of semantic structure. Cambridge: Cambridge University Press. Traugott, E. C. (1982). ‘‘From propositional to textual and expressive meanings: Some semantic-pragmatic aspects of grammaticalization’’, in W. P. Lehman and Y. Malkiel (eds.), Perspectives on historical linguistics. Amsterdam: John Benjamins, 245–271. Traugott, E. C. and R. Dasher (2002). Regularity in semantic change. Cambridge: Cambridge University Press. Turner, K. (1999). The semantics/pragmatics interface from different points of view. Oxford: Elsevier. Visconti, J. (2006). ‘‘The role of lexical semantics in semantic change’’, in M.-B. M. Hansen and K. Turner (eds.), Explorations in the semantics/pragmatics interface. Special issue of Acta Linguistica Hafniensia, 38: 207–234. von Wartburg, W. (1952). Franzo¨sisches etymologisches Wo¨rterbuch (FEW). Basel: Helbing and Lichtenhahn.
Current Trends in Diachronic Semantics and Pragmatics Edited by Maj-Britt Mosegaard Hansen and Jacqueline Visconti r 2009 by Emerald Group Publishing Limited. All rights reserved.
7 THE IMPORTANCE OF PARADIGMS IN GRAMMATICALISATION: SPANISH DIGRESSIVE MARKERS POR CIERTO AND A PROPO´SITO Maria Estelle´s
1. PARADIGMS Unlike routinisation (Hopper, 1987) and context (Hopper and Traugott, 1993 [2003]) – both traditionally regarded as triggers of grammaticalisation – factors such as discourse traditions (Koch and Oesterreicher, 1990; Kabatek, 2001) or paradigms have mainly been seen as responsible for the spread of change, not as the locus which causes new, innovative changes to happen (with a few exceptions, amongst which is Pons Borderı´a, 2006). This paper highlights the role of paradigms as the cause of grammaticalisation processes. A paradigm establishes a particular kind of relationship among a set of signifiers or signifieds called paradigmatic (or associative, according to de Saussure, 1983). Since the elements analysed in this paper have a semantic–pragmatic feature in common that mutually excludes them in some contexts, they will be considered as belonging to one and the same paradigm. Each member, as will be shown, can belong to several paradigms simultaneously: for instance, ciertamente is a member of a semantic–pragmatic epistemic paradigm (when considered in contrast with members such as por cierto or certas, all of them meaning ‘certainly’; for more information see Section 2.2.1.), as well as a member of a formal, morphosyntactic paradigm integrated by adverbs ending in -mente (such as mutuamente ‘mutually’, constantemente ‘constantly’ or lentamente ‘slowly’). Paradigms are the locus where analogy occurs. Analogy is traditionally considered as a modelbased, non-innovative change. This trait differentiates analogy from grammaticalisation, for
124
Maria Estelle´s
instance. However, Kiparsky (2005) maintains that there is an interaction between grammaticalisation and analogy that is ‘‘if anything, the norm, in the sense that most grammaticalisation processes are constrained by, and partly motivated by, the grammatical structure of the language’’. This will be the perspective in this paper. The pressure of paradigms motivates not only the spread of changes, but is also its cause on some occasions: paradigms change; former elements disappear as new ones fill their blank spaces in what Martinet (1955) called ‘‘drag and pull chains’’. And it is likely that new members entering a paradigm do not just adopt a form or become a virtual possibility of permutation in a context; they will possibly take a new path of evolution from the new paradigm into which they are integrated. An attempt to prove this last hypothesis will be offered in next sections by outlining the evolution of a particular set of items belonging to a so-called digressive paradigm in Spanish.
2. THE BIRTH OF A NEW PARADIGM: SO-CALLED DIGRESSIVE MARKERS IN SPANISH Paradigms are continuously changing. This constant change is present in all kinds of paradigms, including paradigms of discourse markers (for a definition of discourse marker, see Martı´n Zorraquino and Portole´s La´zaro, 1999; Portole´s La´zaro, 1998). To show how internal movements and renewals in paradigms provoke atypical evolutions in its members to occur, this paper will select a particular paradigm in Spanish: so-called digressive markers (Casado Velarde, 1998; Portole´s La´zaro, 1998; Martı´n Zorraquino and Portole´s La´zaro, 1999). Although not all experts agree in the list of components in the paradigm, four of them are quite generally regarded as prototypical (Martı´n Zorraquino and Portole´s La´zaro, 1999: 4051–4213), namely por cierto, a propo´sito, a todo esto/a todas estas and dicho sea de paso.1 These elements, although belonging to the pragmatic category of DMs, are very different in origin and thus have a similarly diverse development. Particular attention will be paid to two of these markers: por cierto and a propo´sito. Their development and mutual interaction suffices to provide an overview of the paradigmatic movements and their consequences for grammaticalisation (for an overview including a todo esto/ a todas estas and dicho sea de paso, see Estelle´s, 2009a). The selection of a propo´sito and por cierto has been made for two main reasons: in the first place, as Table 7.1 shows, por cierto and a propo´sito are the most frequently used DM of digression in Spanish; in the second place, they are
1
Spanish tradition in Linguistics considers all four DMs as digressive markers. For the sake of brevity, this paper will maintain the term for it immediately leads to the four markers under discussion. For further discussion on the concept of digressive marker see Pons and Estelle´s (2009).
Spanish digressive markers por cierto and a propo´sito
125
Table 7.1. Frequency of use in digressive markers (CREA corpus) Por cierto
3254 cases – 62%
A propo´sito
1459 cases – 29.5%
A todo esto
274 cases – 5.2%
A todas estas
103 cases – 1.9%
Dicho sea de paso
162 cases – 3.08%
Table 7.2. Circular meaning of por cierto and a propo´sito (DRAE)a por cierto ( . . . ) 2. loc. adv. A propo´sito, viniendo al caso de lo que se dice. (RAE, 2001: XXX) a propo´sito ( . . . ) Y, como conector oracional, ‘por cierto, viniendo al caso de lo que se dice o sucede’: )Me encanta ver bailar. . . A propo´sito, ¿que´ tal le fue en el baile?* (Diccionario panhispa´nico de dudas 2005: 532–533 a
Diccionario de la Real Academia Espan˜ola (DRAE; RAE, 2001) and Diccionario Panhispa´nico de Dudas (DPD; RAE, 2005) are normative dictionaries of Spanish.
the only digressive markers established and recognised as such by normative dictionaries of Spanish; in fact, their respective definitions are circular, as Table 7.2 shows. Using the data provided by electronic corpora of the Royal Academy (diachronic CORDE2 and synchronic CREA3), this article will briefly sketch the particular histories of por cierto and a propo´sito. Afterwards, particular attention will be paid to the moment when their historical pathways merge, as well as to the consequences this junction had for both markers. Therefore, Section 2.1 will be devoted to the evolution of a propo´sito; Section 2.2 will deal with the historical development of por cierto; Section 2.3 will focus on the development of por cierto and a propo´sito altogether; and finally Section 3 will summarise the conclusions drawn in previous sections.
2 The Royal Academy diachronic corpus of Spanish (CORDE). Available online at http://corpus.rae.es/ cordenet.html 3 The Royal Academy corpus of present-day Spanish (CREA). Available online at http://corpus.rae.es/ creanet.html
126
Maria Estelle´s
Table 7.3. Meanings of ‘a propo´sito’ versus meanings of ‘propo´sito’ (RAE, 2001) A propo´sito
propo´sito
1. Adj. phrase
Used to express that something is adequate or appropriate to the aim it was thought of.
‘Aim, goal’
2. Adv. phrase
(Same as de propo´sito) with a specific intention, deliberately and voluntarily.
‘Intention’
3. Connective
Used to express that something, when mentioned, suggests or reminds the idea of talking about a different issue: ‘Speaking of [a propo´sito de] degrees, I’m finishing mine next year’.
‘Issue, topic’
4. Prepositional phrase (a propo´sito de)
Regarding, about.
2.1. A propo´sito The discourse marker a propo´sito is the second in importance and in frequency of use in Spanish, as Table 7.1 showed. However, not all the instances recovered from the CREA corpus can be assigned the same digressive meaning: the DRAE and the most recent DPD present a propo´sito as a polysemic element that has up to five different meanings,4 namely: ‘by the way’, ‘on purpose’, ‘adequate’, ‘regarding (sth)’. Each of the meanings of a propo´sito as a compound relates to a corresponding meaning of the noun propo´sito alone. A propo´sito as a DM (number 3 in the table), as well as a propo´sito/Prepositional Phrase (number 4), are related to the meaning ‘topic’ of the independent noun propo´sito (Table 7.3). 2.1.1. Formal, semantic and pragmatic development. When tracing back the history of propo´sito/ ‘topic’, contexts have been looked for where a and propo´sito belonged to the same syntactic structure. Three main dates must be highlighted in this history, for they are the first attestations of several formal changes in a þ propo´sito: a. 14th century: the first example in Spanish where a and propo´sito5 appear in the same syntactic structure (1); in later examples, as well as in this one, two formal features are recurrent: on the one hand, inserted elements are often found between a and 4
The noun a propo´sito ‘brief piece of theatre’ will be left aside. As we have mentioned above, we refer to the meaning ‘issue, topic’. This is what we will mean by ‘propo´sito’ from now on except as otherwise specifically mentioned. 5
Spanish digressive markers por cierto and a propo´sito
127
propo´sito.6 On the second hand, the structure containing a and propo´sito is usually followed by a PP headed by the preposition de. (1) barruntolo vrias & ouo mjedo & fuyo & fuesse a egipto. & enbio el rrey johaqujm omnes a egipto es de ssaber a elnathan fijo de achbor & otros conel a egipto que lo sacassen de egipto & aduxieronlo al rrey johaqujm & matolo a cuchillo & lano el ssu cuerpo enel fossario de la gente comun. a nuestro proposito de jeremjas pero la uoluntad de achjmini fijo de ssaphan fue buena con jeremjas afin que le non pusiese en poder del pueblo que lo matassen..xxvij. // En prinipio del rreynado de johaqujm fijo de josias rrey de juda fue esta palabra de adona a jeremjas diziendo assy me dize adonay faz (Anonymous work, Biblia romanceada, 14th century). ‘[but when Urijah heard it, he was afraid, and fled, and went into Egypt; And Jehoiakim the king sent men into Egypt, namely, Elnathan the son of Achbor, and certain men with him into Egypt. And they fetched forth Urijah out of Egypt, and brought him unto Jehoiakim the king; who slew him with the sword, and cast his dead body into the graves of the common people. On our topic of Jeremiah: Nevertheless the hand of Ahikam the son of Shaphan was with Jeremiah, that they should not give him into the hand of the people to put him to death.]’7 b. 1400–1411: the first case of a þ propo´sito without any elements inserted in between (2). In many cases like this, the a þ propo´sito structure depends on a verb of movement venir ‘to come’. (2) Buena gente, yo tengo de predicar e dar conplimiento a la materia ayer comenc- ada del general joyzio. E ayer declare´ la ordenac- io´n general e agora declarare´ la definic- io´n sentenc- ial, co´mo sera´n difinidos por sentenc- ia los buenos e los malos. ‘‘Latitudo, longitudo, sublimitas, et profundum’’ (libro et capitulo sicut dixi). Esta palabra propuesta quiere dezir: ‘Anchura, longura, altura, fondura’. Paresc- e que esta palabra non venga a propo´sito, mas sotilmente considerada sı´ faze (San Vicente Ferrer, Sermones, 1411–1412). ‘Good people, I must preach and finish the topic of the general judgement I started yesterday. And yesterday I declared the general ordination, and now the sentential definition, how the just and the evil will be defined by sentence.
6
Between them, mainly definite articles, as well as possessive and demonstrative adjectives, are placed (see Estelle´s, 2009a). 7 The English translation has been taken from King James’s version (Jeremiah 26: 21–24). Only the fragment in bold italics has been added.
128
Maria Estelle´s
‘‘Latitudo, longitudo, sublimitas, et profundum’’ (libro et capitulo sicut dixi). This parable suggested means: ‘width, length, height, depth’. It seems like it is beside the point [lit. ‘not comes to issue’] but, sharply considered, it is not.’ c. 1450: first attestation of a propo´sito with the definitive form of present day Spanish DM a propo´sito (3), without any elements inserted, nor an introductory verb, and lacking the preposition de. (3) Pues, por Dios, mis singulares sen˜ores, usad la fe, guardad la verdad, e no con fingido e simulado amor assı´ entre vos como princ- ipalmente ac- erca de la repu´blica, e considerad que´ valdrı´a un cuerpo humano si oviese sana la mano aviendo enferma la cabec- a e las otras partes del cuerpo. A propo´sito, ¿que´ vale la potenc- ia e riquezas de un c- ibdadano si la patria suya tiene en mala disposic- io´n? (Anonymous writer, Cuatro oraciones a la Repu´blica de Florencia, 1450). ‘For God’s sake, my dear gentlemen, use your faith, keep the truth, and not with artful and pretended love for yourselves and, principally, for the republic, and consider what would a human body be worth if his left hand were in good health whereas his head were sick, as well as the rest of his body. According to this, what is a citizen’s power and wealth worth, if their country is not well?’ The state of affairs provided by examples (1)–(3) is quite unsurprising in the evolution of a DM, considering Traugott (1995): two formerly independent items, a and propo´sito are used in free combination (as in (1), where elements can be inserted in between) and frequently repeated in particular contexts. This leads to fixation and, eventually, attrition (Lehmann, 1982 [2002]; also erosion, according Heine and Reh, 1984), a first step of which is witnessed in example (2), where a propo´sito has no elements inserted, despite the fact that a definite article would have been natural before the noun. It is important to note, however, that the formal differences in (1) and (2) (a þ propo´sito vs. a þ (X) þ propo´sito) relate to a small structural difference: whereas (2) refers to a topic made explicit immediately before, example (1) refers to information given some verses before. To see it more clearly, we provide a wider context of the English translation of the Bible. Here, the main story (Jeremiah’s prophecy in front of an assembly) embeds a parliament of an old man, indented:8 (1b)
8
12: Then spake Jeremiah unto all the princes and to all the people, saying, The LORD sent me to prophesy against this house and against this city all the words that ye have heard. ( . . . )
My emphasis.
Spanish digressive markers por cierto and a propo´sito
129
17: Then rose up certain of the elders of the land, and spake to all the assembly of the people, saying, 18: Micah the Morasthite prophesied in the days of Hezekiah king of Judah, and spake to all the people of Judah, saying, Thus saith the LORD of hosts; Zion shall be plowed like a field, and Jerusalem shall become heaps, and the mountains of the house as the high places of a forest. 19: Did Hezekiah king of Judah and all Judah put him at all to death? did he not fear the LORD, and besought the LORD, and the LORD repented him of the evil which he had pronounced against them? Thus might we procure great evil against our souls. 20: And there was also a man that prophesied in the name of the LORD, Urijah the son of Shemaiah of Kirjath-jearim, who prophesied against this city and against this land according to all the words of Jeremiah: 21: And when Jehoiakim the king, with all his mighty men, and all the princes, heard his words, the king sought to put him to death: but when Urijah heard it, he was afraid, and fled, and went into Egypt; 22: And Jehoiakim the king sent men into Egypt, namely, Elnathan the son of Achbor, and certain men with him into Egypt. 23: And they fetched forth Urijah out of Egypt, and brought him unto Jehoiakim the king; who slew him with the sword, and cast his dead body into the graves of the common people. 24 [absent: on our topic of Jeremiah]: Nevertheless the hand of Ahikam the son of Shaphan was with Jeremiah, that they should not give him into the hand of the people to put him to death. Note, however, that the English translation lacks the fragment containing a þ propo´sito. This phrase appeared in the Spanish version just before verse 24: the author of the Spanish translation adds ‘on our topic of Jeremiah’ to help readers understand that verse 24 does not belong to the old man’s speech anymore. In doing so, he reduces ambiguity and facilitates comprehension; still, he would have had the option of omitting the explicit topic, just as he did with the direction of movement (he does not say ‘back’ nor use specific verbs). This freedom of choice exists, but in (2), guessing which topic the author is referring to is much easier than in (1). This difference is interesting for the development of a propo´sito: whenever a propo´sito appears, the reader must be able to retrieve what ‘topic’ the DM is referring to; the writer can decide to which extent he wants this topic to be explicit: as in (1), it can be to a great extent (by using de followed by the explicit topic) or just schematically explicit (by using verbs like venir or tornar, which respectively inform about the placement of the topic, right before or some lines/ paragraphs before, like in example (2)). But there are also cases like (3), where the author is not explicit at all: example (3), from 1450, is the first attestation that lacks inserted elements, introductory verbs or a preposition de at the end. But in examples like (3), where all these elements are omitted, some restrictions operate that were not necessary in (1) and (2): a propo´sito
130
Maria Estelle´s
relates the forthcoming fragment only to contiguous topics. Otherwise, some mention (with de or a verb of movement) must be made. This topical contiguity does not need to be only textual. Consider example (4). Here a propo´sito could not have been used without specification since it would not link the forthcoming discourse to a contiguous topic, neither contiguous in place (the topic is not in the previous text) nor in time (it is narrated in past tense). The conversation that acts as a frame in (4) is performed by two people who are different from the writer, and in a different time. Therefore, the reader cannot recover the content that explains the cannon shot: it was present in the conversation between the Captain and the King. Had this clue not been provided (that they were talking about war against the Caribs), the fact of shooting a cannonball or talking about two armies’ strength in a farewell would have seemed irreverent or unjustified. (4)
Mie´rcoles, 2 de Enero Salio´ de man˜ana en tierra para se despedir del rey Guacanagari e partirse en el nombre del Sen˜or, e diole una camisa suya, y mostro´le la fuerc- a que tenı´an y effecto que hazı´an las lombardas, por lo cual mando´ armar una y tirar al costado de la nao que estava en tierra, porque vino a propo´sito de pla´tica sobre los caribes, con quien tienen guerra (Ano´nimo, Diarios del viaje de Colo´n, 1492–1493). ‘Wednesday, January 2 He landed in the morning to say goodbye to king Guacanagari and to leave in the name of God, and he gave him a shirt of his, and showed him how strong they were and the effect the cannons caused, so he ordered to charge one of them and to shoot next to the ship anchored in land, because they were talking about the Caribs, whom they were at war against.’
Here we find the root of the first decategorialisation (Hopper, 1991; Brinton and Traugott, 2005). Examples where the preposition de was compulsory after a þ propo´sito gave rise to the PP a propo´sito de: being so attached to and dependent on an intra-sentential element, it became a preposition (functionally speaking). On the other hand, the independence of fragments where a þ propo´sito had no syntactic restriction allowed the combination to reach the extra-sentential level and to become a connective. Still, for these categories to develop, a reanalysis had to take place: (a) (b)
PP: [a [ propo´sito [de Y]]]W[a propo´sito de [Y]] Connective: [a [propo´sito]] YW[a propo´sito] Y
The connective a propo´sito evolves independently and gains abstractness step by step. Chronologically, this evolution comprises three stages: 1. Related to the previous general topic: (3) at first, the meaning of a propo´sito mentioned before – that is, ‘the forthcoming fragment is related to the contiguous topic’ – is indubitable.9 9 An illustrative example is provided by a 16th century proverb, a propo´sito, fray Jarro; by means of this idiom, a speaker whose last words did not follow the thematic line set by the previous discourse, either his or hers (in the same turn of conversation or in monological texts) or some other conversationalist’s discourse, was condemned.
Spanish digressive markers por cierto and a propo´sito
131
2. Related in some way to a particular aspect of the previous topic or situation: from a particular moment in history – around 1700 –, some cases such as (5) appear, which provide evidence of a further step. The fragment with a propo´sito leaves the narration of a daughter being given in matrimony behind and shifts to business. A common ground can be seen in the letters. It could be posited, therefore, that the idea of ‘letters’ triggered the thought (and utterance) of the lack of mail on the business of ‘La Concepcio´n’. This is but a particular point of the previous topic, not the whole of it (the wedding, the daughter): some time before, a de-phrase would have been obligatory. This text, a personal letter written by the king of Spain in 1660, shows a propo´sito already functioning as a connective, and meaning exactly what the dictionary says, that is, ‘suggested by what has been said or by what has happened’: (5)
Al fin de los tres dı´as que nos vimos llego´ el plazo de entregarles a mi hija; ( . . . ) He tenido despue´s cartas suyas en que me dice iban caminando la vuelta de Parı´s buenos, y que ella quiere ya mucho a su marido, de que yo estoy harto gustoso; vuestra carta se la enviare´ en la primera ocasio´n y vos podre´is escribirla cuando os pareciere. A propo´sito, sobre el negocio de la Concepcio´n ha dı´as que no tengo cartas, pero en las postreras que recibı´ me dan muy pocas esperanzas (King Philip IV of Spain, Letter written on the 6th of July, 1660). ‘After the three days we spent together, the time came when I had to give my daughter to them ( . . . ) I have later had letters from her in which she explains to me that they were in good health on their way back from Paris, and that she already loves her husband a lot, which greatly satisfies me; I will send them your letter as soon as I can and you will be able to write to her when you consider it convenient. By the way, on the business of La Concepcio´n, I have not had any letters for days, but the last ones I received do not give me much hope’.
3. Unrelated to the previous topic or situation: an extreme example (6) can be observed in present-day Spanish (1978). Here, the fragment introduced by a propo´sito makes explicit the speaker’s intention in producing his utterance: to avoid silence. In other words, he just picks up the first topic that comes to his mind and speaks up. The fragment selected – Were you in Madrid when your aunt died? Would you see her often? – has nothing to do, either with the situation (a strange feeling about his hand being still upon the girl’s body) or with the last words uttered (the girl’s question about his lack of sleep):
(6)
De pronto vio a Mariana saltar desnuda de la cama [ . . . ] y la sangre se le helo´ en las venas. Por supuesto no fue eso lo que le turbo´, aquella natural falta de pudor, sino el detalle extraordinario de que su mano exploradora, aun cuando la muchacha ya no estaba a su alcance, seguı´a en contacto con su pecho bajo la sa´bana . . . Tuvo la pavorosa sensacio´n de que le habı´an arrancado la mano y de que esa mano seguı´a asida al cuerpo de su sobrina, allı´ de pie a dos metros de distancia.
132
Maria Estelle´s
-¿Que´ pasa, tı´o, ya no tienes suen˜o? En medio del ve´rtigo, replego´ la indecorosa garra ( . . . ), la hundio´ en el bolsillo del batı´n y se incorporo´ como un sona´mbulo. Bien, se dijo, e´ste es el fin de mi titubeante carrera de viejo depravado. -A propo´sito -balbuceo´ por no estar callado, encamina´ndose hacia la puerta con un sudor frı´o en la espalda-. ¿Estabas en Madrid cuando murio´ tu tı´a? ¿La veı´as con frecuencia? (Juan Marse´. La muchacha de las bragas de oro, 1978). ‘Suddenly, he saw Mariana jumping naked out of bed [ . . . ] and he felt petrified. Of course it was not because of that that he became baffled, it was not her natural lack of decency, but the extraordinary fact that his hand, even though the girl was not lying by him anymore, could still feel the contact with her breasts under the sheets . . . He had the impression that his hand had been cut off and that it was still seizing his niece’s body, standing there, two meters in the distance. -What is wrong, uncle? Aren’t you tired anymore? Surrounded by a feeling of lightheadedness, he moved his lewd paw away, sank it into his nightgown pocket and stood up, sleepwalking. Well – he thought – this is the end of my fluctuating career as an old pervert. -A propo´sito (by the way) – he stammered to avoid silence, heading the door with his back covered in cold sweat –. Were you in Madrid when your aunt died? Would you see her often?’ This scale of increasing unrelatedness can be explained by the above-mentioned increase in syntactic freedom of a þ propo´sito. This freedom allows a propo´sito to become relatively independent from the topic to which it refers, and this independence increases with time: after the syntactic one, a semantic–pragmatic rebracketing process has occurred: [[FRAGMENT 1 a propo´sito] FRAGMENT 2]W[FRAGMENT 1 [a propo´sito FRAGMENT 2] The meaning of a propo´sito in examples (1)–(5), encoding something like ‘the forthcoming discourse is related to the contiguous topic’, can be split into two different parts:
There is some kind of relation between this information and the contiguous topic.
There is some discourse coming forth.
Although both were present in a propo´sito, and still are, the pragmatic weight of each one varied through time. If the former examples are regarded in terms of the pragmatic weight given to (a) or (b), it can be seen that
Before the 17th century, the relation (that is, b) was essential.
Between that period and during the 18th century, as in (5), some examples arise where the pragmatic meaning of ‘related to the contiguous topic’ is not as evident as it used to be, although it may not have disappeared.
Spanish digressive markers por cierto and a propo´sito
133
Finally, from the late 1700s on, a propo´sito spreads heading fragments that are fully independent from the previous discourse, and even from the ongoing situation. Thus, in examples like (6), from the 20th century, the instruction given by a propo´sito is more like ‘there is some new discourse coming forth’.
Taking this chronology into account, it can be said that, between the late 1700s and the 1800s, a propo´sito and por cierto start merging their trajectories, for ‘there is some new discourse coming forth’ is almost the same instruction as that encoded by por cierto at that time, and will eventually be the instruction shared by all the members of the paradigm. To illustrate the merging, it will be necessary to first take a look at the evolution of por cierto in Spanish. Section 2.2 will study the DM por cierto until it reaches the stage of evolution when it merges its trajectory with a propo´sito. Afterwards, Section 2.3 will reconstruct the last stage of the development, when a propo´sito and por cierto are already members of the same paradigm. 2.2. Por cierto Given its high frequency and its positional freedom, por cierto is the unmarked member of the paradigm: according to Table 7.1, the Spanish discourse marker por cierto is the most frequently used DM of digression in Spanish (see Mateo Rodrı´guez, 1996; Acı´n Villa, 2000; Reig Alamillo, 2007). Unlike a propo´sito, it does not show much polysemy: the DRAE registers only two meanings (see Table 7.2): por cierto 1. loc. adv. Certainly, in truth. (a) 2. loc. adv. By the way, relevant for what is said. (b) Differently from a propo´sito as well, the frequency of use of each meaning is not balanced at all: more than 95% of the examples of por cierto in present day Spanish instantiate the digressive meaning (b), the CREA corpus shows. por cierto undergoes two grammaticalisation processes. The first one leads an adverbial phrase (7) at the intra-sentential level to become an epistemic DM (8), as in the first meaning of the dictionary (see Estelle´s, 2009b); the second grammaticalisation makes a digressive (9) out of the epistemic structure, and thus gives rise to the second definition of por cierto in the DRAE: (7)
(8)
Agora dime que bestia puede esta seer . . . si non ten por cierto que perdida as la uida (Alfonso X, General Estoria, segunda parte, c.1275). ‘Now tell me what beast this could be . . . otherwise, be certain that your life is lost’. Estas cosas dize el sennor. Si el mio paramiento yo non pus entrel dia & la noche. & non pus Leys al cielo. & a la tierra ( . . . ), por cierto echare yo el linnage de Jacob. & de dauid mio sieruo (Alfonso X el Sabio, General Estoria, parte IV).
134
Maria Estelle´s
(9)
‘Thus said the LORD; If my covenant be not with day and night, and if I have not appointed the ordinances of heaven and earth; Then for sure will I cast away the seed of Jacob, and David my servant.’ cogieron 8, aunque otros dicen fueron 16, y siguiendo el alcance, encontraron con unos moros cosarios, prendie´ndolos a´ todos, sin escaparse hombre . . . . Valiente gente, por cierto, y arriscada, no conociendo al temor, aventurando a´ cada paso la vida (Jero´nimo de Barrionuevo, Avisos. Tomos I, II, III y IV, 1654 – 1658). ‘They captured eight of them, although others say they were sixteen, and, going on with the attack, they found some Moorish pirates, capturing all of them, no man escaping . . . Brave people, indeed/by the way, and daring too, who know no fear and risk their lives in every step they take’.
This section will be divided according to the two different stages of por cierto; in Section 2.2.1 the paradigm of epistemicity will be looked at; in Section 2.2.2 the transition and establishing of the digressive meaning will be analysed. 2.2.1. Epistemic por cierto: First paradigmatic influences. The preposition por and the adjective cierto (‘certain’) are combined and fixed until a new, epistemic DM por cierto is created around 1450.10 The particular details for this evolution will not be given in this article (for more information on this point, see Estelle´s, 2009b). What must be highlighted instead is the internal movement that takes place in the paradigm of epistemicity to which por cierto belongs. A closer look at the DM of epistemicity until the 1500s (the moment in which one of the elements disappears) reveals an interesting subgroup consisting of three elements, all of them derived from the same lexeme: (c/c- )ertas, por cierto and ciertamente. According to Table 7.4 certas was the first particle attested, followed by por cierto and ciertamente. Table 7.5 shows the evolution and distribution of each element until 1500. The data suggest a general tendency towards the spread of por cierto and ciertamente, as well as a gradual reduction in the number of attestations of certas. Until 1250, certas was the unmarked form to express certainty. However, from 1250 on, por cierto assumes this role and shows the highest frequency. Around 1450–1500, only two cases of certas are attested: this old form, which shows the adverbial ending -s and the Latin lexeme cert(without the Spanish diphthong of the stressed eˇWie) will be replaced by the analytic form por cierto, fully Romanic. However, since the 18th century, the preferred form to express certainty in Spanish was ciertamente (Figure 7.1), a learned word from Latin, created with the feminine form
10 Not in such a simple way, as Estelle´s (2009b) posits: all the possible degrees of fixation already coexist from the written origins of Spanish. There was not a period where only adverbial phrases could be found: there could have been a por cierto/DM right from the start.
Spanish digressive markers por cierto and a propo´sito
135
Table 7.4. First attestations (CORDE) Certas
c. 1180
Por cierto
c. 1200
Ciertamente
c. 1250
Table 7.5. Attestations of ‘por cierto’, ‘certas’ and ‘ciertamente’ until 1500 (CORDE) h. 1250
1251– 1300
1301– 1350
1351– 1400
1401– 1450
Certas
56
54
126
272
33
2
Por cierto
18
258
232
321
770
1860
4
138
90
130
532
604
Ciertamente
1451– 1500
3500 Por cierto Ciertamente
3000 2500 2000 1500 1000 500 0 S XIII
S.XIV
S.XV
S.XVI
S.XVII
S.XVIII
S.XIX
S.XX
Figure 7.1. Relative statistics por cierto versus ‘Ciertamente’ (CORDE).
of the adjective, cierta, plus the suffix –mente (oabl. MENTE ‘mind, mood’). This mechanism is the most productive way to create adverbs out of adjectives already in the Renaissance period. The data in Tables 7.4 and 7.5 and in Figure 7.1 can be read in paradigmatic terms: the unmarked member of the paradigm was certas, until it started competing with a new member,
136
Maria Estelle´s
por cierto. From some moment between the late 1300s and the early 1400s, por cierto takes the lead at the expense of certas, assuming all of its formerly exclusive distributions (see Estelle´s, 2009a). Certas practically disappears in the 1500s. This is the first pull-chain (in Martinet’s, 1955 terms) produced in the history of the marker: por cierto replaces certas as the unmarked epistemic. Nevertheless, some years later por cierto is replaced by ciertamente, a form that enters the semantic paradigm of epistemicity, but which formally belongs to a paradigm of adverbs formed by adding -mente, the more productive way of creating adverbs in Spanish, as it was mentioned above. The productivity and quantitative weight of these adverbs played a major role in por cierto being gradually exiled to the periphery of its paradigm. This factor, together with others, enables the transformation of por cierto into a digressive marker. This last step will be investigated in Section 2.2.2. 2.2.2. Digressive por cierto. The epistemic DM por cierto gradually becomes a digressive marker (see Estelle´s, 2009a; Estelle´s, forthcoming). Between the 16th and the 18th century a transitional period occurs, where many cases of por cierto can be read both as epistemic and as digressive (9). After that date, corpora show the first attestations where digression is the only possible interpretation of por cierto. One of the first examples where the digressive interpretation can hardly be called into question is (10) (9)
cogieron 8, aunque otros dicen fueron 16, y siguiendo el alcance, encontraron con unos moros cosarios, prendie´ndolos a´ todos, sin escaparse hombre . . . . Valiente gente, por cierto, y arriscada, no conociendo al temor, aventurando a´ cada paso la vida (Jero´nimo de Barrionuevo, Avisos. Tomos I, II, III y IV, 1654 – 1658). ‘They captured eight of them, although others say they were sixteen and, going on with the attack, they found some Moorish pirates, capturing all of them, no man escaping . . . Brave people, by the way, and daring too, who know no fear and risk their lives in every step they take.’ (10) Aparte de estos almacenes [ . . . ] de Roma, veo otros muchos, abundantemente provistos de los comestibles, ropas, muebles y dema´s efectos que son comunes en toda Europa . . . Por cierto que he entrado en uno a comprar cuerdas de arpa, creyendo haberlas visto en un aparador, y me he encontrado con que allı´ no se vendı´a otra cosa que fideos . . . de diferentes gruesos, eso sı´ . . . -y de aquı´ mi lamentable error, que ha hecho reı´r mucho a toda una prole romana (Pedro Antonio de Alarco´n, De Madrid a Na´poles pasando por Parı´s, el Mont-blanc . . . , 1861). ‘Apart from these stores [ . . . ] in Rome, I see many others, abundantly provided with food, clothes, furniture and other goods that are common in Europe . . . By the way, I went into one of these to buy some harp strings I thought I had seen in a shop window, but I found that they only sold fine noodles . . . of various thicknesses, I must say . . . and that’s why I was led astray, which made a whole Roman crowd laugh a lot’. (Estelle´s, 2009a)
Spanish digressive markers por cierto and a propo´sito
137
This change between por cierto in (9) and in (10) is caused by a combination of two factors: the generalisation of por cierto in semantically – and syntactically – subsidiary constructions and the particular kind of epistemicity conveyed by por cierto. In the following lines, both aspects will be developed: the syntactic contexts in which the epistemic por cierto appears, and how por cierto acquires this new meaning (due to the role of the non-focal epistemicity of por cierto). 2.2.3. The origin of the digressive meaning. Where does the meaning of digression come from? Unlike a propo´sito, it is difficult to see the root of the change in the meaning of por cierto. In a propo´sito, a link between a meaning ‘aim, goal’, ‘intention’ and ‘topic’ is evident: what is a ‘topic’ but the aim or the intention of a text? In por cierto, however, it is hard to envisage an intersection in the meanings of ‘certainty’ and ‘digression’ and, even if it could be thought of, it would be a really forced one; the data show until the 17th–18th century, por cierto appears mainly in appositions and exclamations. The first group refers back to the previous utterance and characterises it, either the whole of it or simply a part (9), and the second group includes exclamations (11) in which the author evaluates the previous discourse (and sometimes the following discourse): (11)
E´sta era la razo´n comu´n; empero sobre e´sta se alegraban interiormente en la consideracio´n de que por las conveniencias del estado de Francia fuesen tan propicios los accidentes de Espan˜a, que ningu´n juicio dejarı´a de abrazar sus intereces; que era preciso el echar mano de las turbaciones del enemigo como de materiales utilı´simos para la serenidad propria. ¡Miserable condicio´n, por cierto, de la fortuna, que no tiene caudal para fabricar gran imperio a un prı´ncipe, sino con las ruinas de otro! Ası´ resolutos, eligieron entre todos a Francisco Vilaplana, caballero perpin˜ane´s . . . (Francisco Manuel de Melo, Historia de los movimientos, separacio´n y guerra de Catalun˜a, 1645). ‘This was the common reason, but above all they rejoiced in thinking that, considering the interests of France, the disgraces of Spain were so convenient, and no judgement would prevent them to hold those interests because it was necessary to use the enemy’s disturbance as a most useful material to guarantee their one’s calmness. Pitiful condition, indeed (/by the way), of fortune, which has no richness to build a great empire for a prince but uses the ruins of another! With that decision in mind, they elected Francisco Vilaplana, a gentleman from Perpignan . . . ’
These constructions containing por cierto play different syntactic roles. However, in the informational structure, it is possible to find a common feature to all of these functions: they are tangential elements that provide additional, non-essential information, relatively distinct from the information given in the main clause they are interrupting, i.e. they are parenthetical (for more
138
Maria Estelle´s
information, see Dehe´ and Kavalova, 2007). This meaning is not present in any of the words contained in the structures, but rather in the structure itself. In Nuyts’s (2001: 271) words: Quotation syntactic patterns, even if probably not entirely independently from the lexical elements figuring in them, have a symbolic value, too, which they must develop and which must get conventionalized.
This feature of providing additional, side information can be easily seen when the discursive fragment occurs within parentheses or parenthetical marks (12): (12)
Subı´ la montan˜a de Nuestra Sen˜ora de Montserrat (cosa, por cierto, milagrosa y digna de ver), en donde vi maravillas y infinidad de gentes que cada dı´a acuden a la fama, y por sus devociones, votos y promesas (Mateo Luja´n de Saavedra (Juan Martı´) Segunda parte de la vida del pı´caro Guzma´n de Alfarache, 1602). ‘I climbed the mountain of Nuestra Sen˜ora de Montserrat (something, by the way / truly, miraculous and worth seeing), where I saw wonders and crowds who gather every day for devotion, vows and promises, all attracted by its fame’.
To sum up, there is a variety of examples in which por cierto appears in syntactic structures that have in common the fact of being parenthetical and carrying additional information, different from the prior one: every apposition is a new little topic; for that reason, some devices are needed to tell the listeners not to consider what is going to be said next as nonsense: a particular intonation, a pause or a DM as por cierto indicate to readers or listeners that the fragment they introduce is worth processing, that it is not a red herring. At this point both fields, the epistemicity and the digression, seem to converge: due to the frequent use of por cierto in parenthetical syntactic structures, which carry the secondary metalinguistic implicature of conveying additional information, por cierto would acquire this meaning of ‘new relevant information’ so far only present in the meta-discursive level. It is likely that this new notion became a part of its meaning (Estelle´s, 2009a; Estelle´s, forthcoming). This explanation provides the source of the new meaning: the informational structure of parenthetical syntactic structures. But the question arises as to what happened with the epistemic meaning and why it was lost. Grammaticalisation studies account for the loss and further change of meaning in terms of a loss of semantic features (see, for instance, Lehmann, 1982 [2002]) or a change in the domain, where only the schematic image is preserved (Sweetser, 1988). The data favour the latter interpretation: the epistemic PC, when placed in an embedded, subordinated structure, does not preserve fully epistemicity as it would do in a main clause. Instead, it becomes a focal particle qualifying (mainly intensifying) the structure it is inserted into, rather than expressing certainty about a state of affairs. This emptying of conceptual meaning is not
Spanish digressive markers por cierto and a propo´sito
139
diachronic, but synchronic, and is due to changes in domain. To illustrate this synchronic process, see ciertamente in different environments in present day Spanish (a–c): a. Esto que te cuento, lo se´ certamente. ‘What I’m telling you, I know it for sure’. b. Ciertamente, Obama sera´ el personaje ma´s importante del an˜o. ‘Certainly, Obama will be the most important person of the year’. c. Ayer la vi con un vestido, ciertamente precioso, que ya le habı´a visto antes. ‘I saw her yesterday, wearing a dress –really beautiful- that I had already seen on her’. In (a)–(c), a gradation can be established. In (a), ciertamente clearly means epistemicity: it expresses certainty about a state of affairs; (b) can also be understood as expressing epistemicity, but not as straightforwardly as (a): the speaker qualifies his/her conviction, rather than the facts themselves; finally, there is few – if any – certainty in (c). In structures like (c), ciertamente could be replaced by other elements such as su´per or la mar de (‘very, very’ or ‘extremely’). It functions, therefore, as a quantitative, subjective valuation, i.e. as an intensifier (Albelda, 2007: 146–155). In terms of Nuyts (2001) the epistemicity of (c), just as the structures it is inserted in, would become non-focal, i.e. it does not carry the main information of the sentence anymore. Instead, other additional factors, independent from epistemic qualification, do play a role. The focus is most clearly reflected in fragments that contrast with other parts of the discourse, as well as in those passages where writers or speakers are forced to express their opinion about the likelihood of a state of affairs. This could happen, for instance, in a dialogical exchange (like (a) or (b)), but also in monological pieces, if the previous text could also invite the author to evaluate the probability of a certain state of things (13) (13)
Senyors jutges lo que panaus dize elo que el qujer ajustar ala ley ultra lo que noy trobamos scripto sufris lo uos no Res & si esto fues que vos lo sufriesedes porel ajutorio que el ha adquesido esto seria contra la dignjdat de vos & contra honorabiljdat Mas cuydades vos quel pueblo lo sufra no por c- ierto c- ierta ment yo conozco el seso & el saber que en [vos] es . . . (Anonymous author. Libro del Tesoro de la Catedral de Girona, 1400–1425). ‘Judges, what Panaus says and what he intends to adjust to the law, going beyond what is written, do you suffer it? Not at all; and if that be the case, that you suffered this due
140
Maria Estelle´s
to the help he’s achieved, this would go against your dignity and your honour. But, do you care if people suffer it? Of course not. Certainly, I know your sense and your wisdom . . . ’ Examples provided above can be classified according to their focal or non-focal nature. Thus, example (13) would be focal, whereas (11) would be non-focal, for por cierto neither contrasts with the previous text nor answers a question of any kind. What is more, its non-focality is supported by the fact that por cierto, as could be seen before, appears in embedded clauses, which carry secondary, backgrounded information. Syntactic subordination appears strongly correlated with informational subordination: main clauses prototypically carry foregrounded and embedded clauses backgrounded information (Nuyts, 2001: 264). It is noticeable that, one by one, all examples in which por cierto can be read as a digressive marker are non-focal cases, whereas examples in which por cierto occupied a focal informational position remained epistemic and gradually disappeared. That is why por cierto, not emptied of epistemic meaning, but rather never having had any in these syntactic environments, can acquire the new digressive meaning that was present in the informational, pragmatic structure of the syntactic construction in which it was constantly repeated from the 16th century onwards. At some moment between the 17th and the 18th century, a propo´sito and por cierto convey the same meaning, ‘new relevant information’ (see Pons Borderı´a and Estelle´s, 2009). The only difference is that they appear in different syntactic environments. 2.3. Por cierto and a propo´sito. Interaction and mutual consequences In sum, two markers, a propo´sito and por cierto, share the same procedural meaning but appear in complementary distribution: 1. a propo´sito as a DM appears heading paragraphs and whole sentences, structures that formally reproduce the newness of the information it conveys. This is the case of (5), reproduced again below, in which a propo´sito retrieves a secondary part, barely a mentioned word, and starts a new paragraph that develops it. (5)
Al fin de los tres dı´as que nos vimos llego´ el plazo de entregarles a mi hija; ( . . . ) He tenido despue´s cartas suyas en que me dice iban caminando la vuelta de Parı´s buenos, y que ella quiere ya mucho a su marido, de que yo estoy harto gustoso; vuestra carta se la enviare´ en la primera ocasio´n y vos podre´is escribirla cuando os pareciere. A propo´sito, sobre el negocio de la Concepcio´n ha dı´as que no tengo cartas, pero en las postreras que recibı´ me dan muy pocas esperanzas (King Philip IV of Spain, Letter written on the 6th of July, 1660). ‘After the three days we spent together, the time came when I had to give my daughter to them ( . . . ) I have later had letters from her in which she explains to me that they were in
Spanish digressive markers por cierto and a propo´sito
141
good health on their way back from Paris, and that she already loves her husband a lot, which greatly satisfies me; I will send them your letter as soon as I can and you will be able to write to her when you consider it convenient. By the way, on the business of La Concepcio´n, I have not had any letters for days, but the last ones I received do not give me much hope’. 2. por cierto appears in subordinated contexts, either semantically or syntactically. By subordinated, then, we mean appositions, relative–explicative clauses, exclamations praising a previous idea or simply cases in which por cierto frequently acts as an emphasiser of a single word (it acts as an adjective). Once a new paradigm emerges, an unmarked member must naturally reveal itself. The fight between the markers, as dictionaries and grammars show, was won by por cierto, for the following reasons: 1. por cierto had only one meaning at the time: the epistemic. In turn, a propo´sito was a polysemic expression that derived from a polysemic word – ‘propo´sito’ – as well. The search for the ideal situation one word-one meaning makes por cierto perfect to be a core, unmarked member. 2. In addition, due to the pressure of ciertamente, por cierto was itself being gradually rejected from the core of its paradigm of epistemicity. 3. In nearly one in four cases,11 this epistemicity in por cierto is a peripheral manifestation of an epistemic DM, that is, it does not convey real or full certainty, but rather functions to intensify or emphasise the element(s) it affects. 4. These cases in which por cierto is not focal show the DM in tangential, informationally backgrounded structures. The informational structure of these constructions convey precisely this background implicature: being non-essential, secondary, relatively unrelated to the previous discourse; in one word: new with respect to the topic under discussion. 5. The form por cierto is shorter and more natural in Spanish (generally tending to avoid words stressed in the third-last syllable). Due to these factors, por cierto takes the lead as the unmarked member of this paradigm. And thus, it spreads to all contexts where formerly only a propo´sito was to be found. In the 19th century we find examples like (10), reproduced again below, in an environment peculiar to a propo´sito so
11
The CORDE corpus registers 7524 cases of por cierto between 1500 and 1700. Nearly 25% of them are non-focal cases, placed in subordinated environments as those mentioned above. A propo´sito registers 2864 cases, which must be distributed between four different categories (adj. phrase, prepositional phrase, DM, free construction). That is, before becoming a digressive, non-focal epistemic, por cierto registered a high frequency of appearance.
142
Maria Estelle´s
far, heading a new sentence. This process continues, with por cierto gradually acquiring more of a propo´sito’s exclusive positions (the same process occurred 400 years before, with por cierto assuming the privative distributions of certas). This evolution culminates in the 20th century, where examples of por cierto heading a new speech act appear. Before the 1930s–1940s, corpora do not show any attestations of por cierto like (14) placed before an interrogative sentence: (10)
(14)
Aparte de estos almacenes [ . . . ] de Roma, veo otros muchos, abundantemente provistos de los comestibles, ropas, muebles y dema´s efectos que son comunes en toda Europa . . . Por cierto que he entrado en uno a comprar cuerdas de arpa, creyendo haberlas visto en un aparador, y me he encontrado con que allı´ no se vendı´a otra cosa que fideos . . . de diferentes gruesos, eso sı´ . . . -y de aquı´ mi lamentable error, que ha hecho reı´r mucho a toda una prole romana (Pedro Antonio de Alarco´n, De Madrid a Na´poles pasando por Parı´s, el Mont-blanc . . . , 1861). ‘Apart from these stores [ . . . ] in Rome, I see many others, abundantly provided with food, clothes, furniture and other goods that are common in Europe . . . By the way, I went into one of these to buy some harp strings I thought I had seen in a shop window, but I found that they only sold fine noodles . . . of various thicknesses, I must say . . . and that’s why I was led astray, which made a whole Roman crowd laugh a lot.’ (Estelle´s, 2009a) En este momento me llama Germaine para decirme que le gusta mucho el poema tuyo ‘‘Amiga’’: ‘‘Para cristal te quiero . . . ’’. Es en la antologı´a del sen˜or Prampolini Cosecha -que hoy he traı´do de la biblioteca-: antologı´a preciosa, de la que nosotros no tuvimos la menor referencia en cuanto ‘‘antologizados’’. (Por cierto ¿no esta´ este sen˜or Prampolini en Columbia? ¿No le conoces?) Germaine me decı´a -y termino la narracio´n . . . (Jorge Guille´n, personal letter, 1940). ‘Then Germaine calls me to tell me she loves your poem ‘‘Amiga’’: ‘‘Para cristal te quiero . . . ’’. It is in the anthology of Mr. Prampolini Cosecha – which I brought today from the library-: a beautiful anthology, of which we knew nothing, even though we were included in it (By the way, isn’t this Mr. Prampolini in Columnbia? Do you not know him?) Germaine was telling me- and I finish my narration’ . . .
With the step illustrated in (14), there are not any privative environments of a propo´sito anymore. However, a parallel appropriation of contexts does not occur the other way round: por cierto prevents a propo´sito from accessing any of its exclusive contexts. At least, this was the situation until the 20th century. Again, none of the corpora show a single example where a propo´sito appears in an apposition or subordinate structure. However, the Internet retrieves several hundreds of examples in present day Spanish, like (17): (17)
Varios comentarios de Woolf en ese artı´culo (magnifico, a propo´sito) parecen prefigurar experimentos a la Pynchon (ellamentodeportnoy.blogspot.com/2006/05/la-narrativa-moderna_06.htm).
Spanish digressive markers por cierto and a propo´sito
143
‘Several comments made by Woolf in that article (excellent, by the way) seem to anticipate experiments a` la Pynchon’. This fact is even more significant than por cierto assuming a propo´sito’s contexts: a propo´sito was more advanced in the process of grammaticalisation: its scope was wider than por cierto’s. It could head a sentence, a paragraph and showed quite early a total semantic unrelatedness. por cierto, when entering the paradigm, gradually (but quickly) has access to those more advanced steps already reached by a propo´sito and incorporates them. It could be expected that this was the end of the merging: from then on, both markers would evolve the same way and would keep on gaining new grammatical meanings (Hopper and Traugott, 1993). However, what can be witnessed is a step backwards: a propo´sito reduces its scope (against the expectation for DMs) and, although very recently indeed, it is showing a tendency to appear at the intra-sentential level, in positions that formerly only por cierto would occupy. These positions represent a dramatic internal change for a propo´sito, since it stops being a two-place operator and becomes a one-place operator: it is placed in the functional context of a focal particle, although it has not (yet) developed as such.12
3. SUMMARY Analogy is not just a change experienced by a member that also affects other members: just like raindrops merge their trajectories, different members of the same paradigm end up experiencing quite similar pathways of change that fall in principle out of their predictable evolution (v.gr. unexpected positions, pragmatic restrictions, etc.). In other words, keeping within the boundaries of a particular paradigm causes the components of this paradigm to inherit some of the syntactic and pragmatic characteristics of the rest, and to share their own peculiarities. The Spanish DM a propo´sito derives from constructions in which the word propo´sito referred to a previous topic. The expression was used to justify the relevance of the apparently new fragment introduced: it was somehow related to a previous topic. Through the years, the ‘newness’ of the fragment introduced is gradually foregrounded, whereas the ‘relatedness’ is backgrounded. This process starts at the 17th century and is complete one century later. Meanwhile, an epistemic DM por cierto experiences a banishment from the core of the epistemic paradigm and is completely replaced by ciertamente as the unmarked option in the 18th century. While this process is happening, por cierto is being reanalysed as a digressive marker (16th–18th century) due to (a) its repetition in structures conveying a secondary inference ‘new
12
That is, it appears in subacts. See Pons and Estelle´s (2009).
144
Maria Estelle´s
information’, and (b) the fact that nearly 25% of the attestations of por cierto show an empty marker, a peripheral epistemic that is not conveying real epistemicity and that practically lacks conceptual meaning. This last factor allows por cierto to acquire a new meaning, namely the one that was formerly present in the structures in which por cierto was repeatedly featured (‘new information’). From the 18th century on, both markers start conveying the same information ‘new, relevant information’ (see Pons Borderı´a and Estelle´s, 2009), and both are integrated in a paradigm created anew, a paradigm of digression. por cierto becomes the unmarked member due to its history, its progressive exile from the epistemic paradigm and its advantages over a propo´sito (por cierto is more economical and less polysemic). As the unmarked member, it spreads to all environments where formerly only a propo´sito was to be found (such as heading paragraphs). The corpus of present day Spanish (CREA) seems to evidence that por cierto still monopolises some environments such as appositions or parenthetic constructions. Notwithstanding this fact, searches in the Internet reveal that, in the last decade, some examples are attested in which a propo´sito appears in por cierto’s former exclusive contexts. Therefore, diachrony evidences that belonging to a paradigm causes its members not only to experience punctual changes, but also to adopt and follow new pathways of grammaticalisation, impossible to predict if such paradigmatic associations had not taken place. The pressure of the digressive paradigm explains why por cierto starts having scope over paragraphs, but (what is more important) it also accounts for a more interesting fact: a propo´sito, a former connective, reduces its scope and becomes a one-place operator appearing in the intra-sentential level. Members of the same paradigm can be mutually affected by the form or the meaning of the rest, but they can also acquire their way of evolving, of grammaticalising. This is especially true for prototypes, which irradiate some of their characteristics to other, less prototypical members.
REFERENCES Acı´n Villa, E. (2000). ‘‘‘Por cierto’, ‘a propo´sito’ y otros digresivos’’, in P. Carbonero Cano, M. Casado Velarde and P. Go´mez Manzano (eds.), Lengua y discurso. Estudios dedicados al profesor Vidal Lamı´quiz. Madrid: Arco Libros, 59–72. Albelda, M. (2007). La intensificacio´n como categorı´a pragma´tica: revisio´n y propuesta. Frankfurt am Main: Peter Lang. Brinton, L. J. and E. Traugott (2005). Lexicalization and language change. Cambridge: Cambridge University Press. Casado Velarde, M. (1998). ‘‘Lingu¨´ıstica del texto y marcadores del discurso’’, in M. A. Martı´n Zorraquino and E. Montolı´o Dura´n (eds.), Los marcadores del discurso. Teorı´a y ana´lisis. Madrid: Arco, 55–70. de Saussure, F. (1983). Course in general linguistics. Chicago: Open Court Classics.
Spanish digressive markers por cierto and a propo´sito
145
Dehe´, N. and Y. Kavalova (eds.) (2007). Parentheticals. Amsterdam: John Benjamins. Estelle´s, M. (2009a). Gramaticalizacio´n y gramaticalizaciones. La evolucio´n del paradigma de marcadores del discurso de digresio´n en espan˜ol, PhD Thesis, Universitat de Vale`ncia. Estelle´s, M. (2009b). ‘‘The Spanish discourse marker por cierto: A case of non-standard grammaticalization’’, in A. Spiridon, C. Ricci and C. Rossari (eds.), Grammaticalization and pragmatics: Facts, approaches, theoretical issues, Studies in Pragmatics 5. Bingley: Emerald, 93–114. Estelle´s, M. (Forthcoming). ‘‘Un caso atipico de gramaticalizacion (II). El valor digresor de por cierto’’, Rilce. Heine, B. and M. Reh (1984). Grammaticalization and reanalysis in African languages. Hamburg: Helmut Buske. Hopper, P. (1987). ‘‘Emergent grammar’’. Berkeley Linguistics Society 13: 139–157. Hopper, P. and E. C. Traugott (1993 [2003]). Grammaticalization. Cambridge: Cambridge University Press. Hopper, P. J. (1991). ‘‘On some principles of grammaticalization’’, in E. C. Traugott and B. Heine (eds.), Approaches to grammaticalization Vol. I. Amsterdam: John Benjamins, 17–36. Kabatek, J. (2001). ‘‘¿Co´mo investigar las tradiciones discursivas medievales? El ejemplo de los textos jurı´dicos castellanos’’, in D. Jacob and J. Kabatek (eds.), Lengua medieval y tradiciones discursivas en la Penı´nsula Ibe´rica. Frankfurt: Vervuert, 97–132. Kiparski, P. (2005). Grammaticalization as optimization. (Draft). Koch, P. and W. Oesterreicher (1990). Gesprochene Sprache in der Romania: Franzo¨sisch, Italienisch, Spanisch. Tu¨bingen: Niemeyer. Lehmann, C. (1982 [2002]). Thoughts on grammaticalization. ASSIDUE-Arbeitspapiere des Seminars fu¨r Sprachwissenschaft der Universita¨t Erfurt 9. Available online at http://www.db-thueringen.de/servlets/ DerivateServlet/Derivate-2058/ASSidUE09.pdf Martinet, A. (1955). Economie des changements phone´tiques. Traite´ de phonologie diachronique. Bern: Francke. Martı´n Zorraquino, M. A. and J. Portole´s La´zaro (1999). ‘‘Los marcadores del discurso’’, in I. Bosque and V. Demonte (eds.), Grama´tica descriptiva de la lengua espan˜ola. Madrid: Espasa-Calpe, 4051–4213. Mateo Rodrı´guez, J. E. (1996). ‘‘Los marcadores digresivos. Estudio especial de ‘por cierto’ en espan˜ol actual’’, in M. Casado Velarde (ed.), Scripta philologica in memoriam Manuel Taboada Cid. La Corun˜a: Universidade da Corun˜a 2, 531–552. Nuyts, J. (2001). Epistemic modality, language and conceptualization: A cognitive-pragmatic perspective. Philadelphia: John Benjamins. Pons Borderı´a, S. (2006). ‘‘From pragmatics to semantics: esto es in formulaic expressions’’. Acta Lingvistica Hafniensia 38: 180–206. Pons Borderı´a, S. and M. Estelle´s (2009). ‘‘Expressing digression linguistically. Do digressive markers exist?’’. Journal of Pragmatics 41 (5): 931–936. Portole´s La´zaro, J. (1998). Los marcadores del discurso. Barcelona: Arco. RAE (2001). Diccionario de la lengua espan˜ola (DRAE). Madrid: Espasa-Calpe. RAE (2005). Diccionario panhispa´nico de dudas. Madrid: Espasa-Calpe. Reig Alamillo, A. (2007). ‘‘‘Por cierto’ y la digresio´n’’. Oralia 10: 233–254. Sweetser, E. (1988). ‘‘Grammaticalization and semantic bleaching’’, in S. Axmaker, A. Jaisser and H. Singmaster (eds.), Proceedings of the Fourteenth Annual Meeting of the Berkeley Linguistics Society. Berkeley: Berkeley Linguistics Society, 389–405. Traugott, E. (1995). ‘‘The role of DM in a theory of grammaticalisation’’, in ICHL XII Manchester, Version of 11/97. Available at: http://www.stanford.edu/Btraugott/ect-papersonline.html
Current Trends in Diachronic Semantics and Pragmatics Edited by Maj-Britt Mosegaard Hansen and Jacqueline Visconti r 2009 by Emerald Group Publishing Limited. All rights reserved.
8 THE MULTIPLE ORIGIN OF ES QUE SPANISH: DIACHRONIC EVIDENCE
IN
MODERN
Magdalena Romera
1. INTRODUCTION From a discourse point of view the Spanish expression es que is a functional element that allows relating or interpreting two utterances in a coherent way, that is, es que can be considered what has been traditionally called a discourse marker or what I call here a discourse functional unit (DFU hereafter) (cf. Romera, 2004).1 Most analysts agree that utterances introduced by es que state the cause, reason, explanation, or justification of a previous utterance (Ferna´ndez Leborans, 1992; Espan˜a Villasante, 1996; Fuentes Rodrı´guez, 1997; Romera, 2004 among others). For some other authors, es que can also introduce reformulation and reinterpretation (Delahunty and Gatzkiewicz, 2000)2. So in example (1) below, unit 5 is a justification for the
1
I follow here the terminology used in Romera (2004), according to a theoretical framework of coherence discourse relations (Mann and Thompson, 1988; Hobbs, 1990, 1996; Sanders et al., 1992; Polanyi, 1996, among others). In this model, discourse is structured in significant minimal units which combine through a series of semantic-logical relations. Minimal units can be of two types: propositional units (DPUs) which represent an instantiation of actual events and states of affairs and functional units (DFUs) which transmit conceptual and combinatorial content. Functional units express discourse relations that indicate how to combine different propositional units. 2 The types of relations I propose in this article for es que are also within the category of causal type of relations (cf. Hovy and Scott, 1996). However, as one anonymous reviewer suggests, the assignment of relations is in fact a rather subjective task. All discourse analysts are aware of such a problem and some studies have addressed this issue specifically. Marcu et al. (1999) followed a series of experiments for measuring agreement of discourse structure annotations and suggested that the best method to assure objectivity is annotation by several judges. These authors showed that several discourse annotators can reach a significant agreement (more than 80%) when trained together.
148
Magdalena Romera
request of information performed in 1, where the speaker becomes interested in some clothes which are different from what she initially said she wanted to buy. (1)
S1 S2 S1 S2 S1
1 ¿Y estos?, ¿no son vestidos? o . . . 2 ¿Estos? 3 Sı´. 4 No, osimulta´neoW para ti son muy serios. 5 ¡;Ah! o/simulta´neoW 6 Es que me habı´- . . . me habı´a llamado la atencio´n osimulta´neoW el color.
S1 S2 S1 S2 S1
1 And these?, aren’t they dresses? Or . . . 2 these? 3 yes. 4 no, osimultaneousW they are too serious for you. 5 oh! o/simultaneousW 6 it is that the color had- . . . had caught my eye.
CREA
Like other functional units, from a diachronic point of view, es que seems to have evolved from a structure which initially did not serve a grammatical function. Es que is an expression constructed by a copula (ES ‘is’) and a subordinate conjunction (QUE ‘that’) and can be found in canonical examples like (2): (2)
La primera norma es que todos lleguen puntuales ‘The first rule is that all arrive on time’
The purpose of the present study is to explain how a subordinate structure like the one in (2) evolved in form and meaning into the expression of functional meanings. The data analyzed here suggests that subordinate structures with es que undergo a process of grammaticalization. Semantically, from heading an utterance which states the existence of an event or a state of affairs, es que structures come progressively to introduce elaboration or reinterpretation of the content of a previous utterance. Formally, the elements placed in the subject position gradually lose their referential status and subordinate structures with es que may eventually appear as subjectless structures. Based on the analysis of large corpora, I here present a study on the evolution of es que. I review previous studies on the origin of constructions with es que and suggest a hypothesis of the diachronic evolution of this form. I then present the diachronic analysis of data from the Corpus del Espan˜ol and the findings of this analysis. In the final section, I present my conclusions.
Origin of es que
2. TYPES
OF ES QUE
149
CONSTRUCTIONS
Previous studies claim that es que can appear in so-called inferential constructions, also called sentential focus constructions and sentential focus clefts (cf. Delahunty, 2001; Delahunty and Gatzkiewicz, 2000). They are referred to as ‘inferentials’ (Delahunty, 2001) for they can be paraphrased as ‘‘infer X from Y’’. For instance, a sentence such as ‘‘It’s just that they can’t bear anybody to have an advantage’’ (Delahunty, 2001: 522) can be paraphrased as ‘‘infer from what has just been said that they can’t bear anybody . . . ’’. The studies on Spanish agree on distinguishing two types of syntactic structures with es que (Ferna´ndez Leborans, 1992). The first structure has its subject position3 filled as in la primera norma es que todos lleguen puntuales (the first rule is that all arrive on time). The second does not have a subject and es que seems to introduce an independent syntactic clause, as in es que no encuentro las llaves (it is that I can’t find the keys). In both cases, the predicate position is occupied by a subordinate clause which is semantically focalized. Structurally, these two types of sequences look like: 1 Explicit subject þ COPULA (ES) þ QUE þ Subordinate clause 2 No subject þ COPULA (ES) þ QUE þ Subordinate clause The first type can have different elements as subjects: a semantically definite NP as in (3), an anaphoric pronoun as in (4), an indefinite NP as in (5), and a pseudo-cleft clause introduced by LO QUE as in (6). (3) La primera regla es que los estudiantes sean puntuales ‘The first rule is that students are always on time’ (4) Eso es que no tiene muchas luces ‘That is that he is not very bright (i.e., ‘That means that he is not very bright’)’ (5) El caso es que no nos pudimos marchar ‘The thing is that we could not leave’ (6) Lo que vemos todos los dı´as es que . . . ‘What we see everyday is that . . . ’ I have called this first type of structure integrated constructions, since es que is part of a complete copulative clause.
3
I will not discuss here whether subject and predicate occupy the pre- and post-copula places, respectively. As is accepted in the field, I assume that a copula may take a subject and a predicate in either order (see Partee, 1986, among others). For convenience, I will call the pre-copula place ‘‘the subject position’’ and the post-copula place ‘‘the predicate position’’.
150
Magdalena Romera
The second type of construction is that in which there is no subject of the copula. The following examples belong to this group: (7) Castigadle, si es que ha pecado ‘Punish him, if it is that he has sinned’ (8) Una cosa es cierta, y es que seguiremos adelante ‘One thing is true, and it is that we keep on going’ (9) ¿Te apetece venir a bailar? Es que a mı´ esas cosas no se me dan muy bien ‘Would you like to go dancing?’ ‘It is (i.e., ‘the thing is’) that I am not very good at those things’ I have called these cases nonintegrated constructions, in contrast to the previous ones. The question to answer is whether there is a diachronic relationship between these two types of structures (integrated and nonintegrated), and if so, why and when subjects of the copula disappear. Almost all authors agree that synchronically there is indeed a relationship between the two. Declerck (1984, 1992), Ferna´ndez Leborans (1992), and Espan˜a Villasante (1996) maintain that the presence or the absence of a subject in copulative structures is only an external mark that does not imply changes in the underlying syntactic structure. Subjectless structures with es que are mainly observed in languages with a rich inflectional verbal system such as Spanish and Italian, cf. Espan˜a Villasante (1996: 131). For Ferna´ndez Leborans (1992), the copula in es que constructions does not represent a predicational value, but rather the identification between the subject and the focalized element which occupies the predicate position. When the subject is missing, its reference can be recovered from the previous context or from the communicative situation. A similar thesis is defended by Delahunty (2001) and Delahunty and Gatzkiewicz (2000), who consider that the structures with es que without a subject leave an open referent to be set up by pragmatic knowledge. A different perspective may explain the relationship between the two constructions and shed some light on the evolution of es que. There is semantic identification between the two arguments of the copula (subject ¼ predicate); however, we may also attribute predicate meaning to the copula, especially in nonintegrated es que constructions. For instance, in example (10), it is possible to paraphrase the copula þ que as OCURRE QUE ‘it occurs that’, PASA QUE ‘it happens that’. This will indicate that there is also a predicational relationship between the two arguments. (10) es que no me apetece pasa que no me apetece4 ‘it is (i.e., the thing is) that I don’t feel like it’ ‘it (so) happens that I don’t feel like it’ This can be explained in terms of a continuum between the definitional and predicative meanings or uses of the copula, assuming that each of these meanings is possible depending on 4 Espan˜a Villasante (1996: 130) argues that these sentences are ungrammatical. However, all Spanish native speakers consulted agree that these paraphrases are grammatical in the appropriate context.
Origin of es que
151
the propositional content of the arguments related. In this line of thinking, Hedberg (1990), following Partee’s (1986) analysis, suggests a typology of copular clauses in cleft sentences according to subject and predicate semantic types. In this sense, a subordinate clause as a predicate would argue in favour of a predicational construction. This is confirmed by the fact that most examples of nonintegrated es que constructions can be easily paraphrased with a predicational verb such as PASAR, as we saw above. Also, from a discourse point of view, the predicative meaning of the copula in es que constructions plays a significant role. The es que clause relates to a previous unit not only by a purely definitional relation, but also by elaboration and explanation relations as well, as mentioned before. This would not be possible if the meaning of the copula were only identificational. The fact that copula can be interpreted as definitional and predicational at the same time could show the diachronic evolution of the meanings of es que structures. The thesis defended here is that es que was initially used to express a purely definitional meaning, that is, to present the existence of an event. Then, the expression of this event also represents a more specific elaboration of a previous segment. Finally, the elaboration allows for more subjective content in terms of the speaker’s interpretation. Formally, I suggest that the subject in integrated constructions disappears due to a loss in the referential status of the elements placed in subject position. These elements are initially highly referential and can be related anaphorically to the previous content, but later they are mere anticipatory elements that introduce the focalized subordinate clause. Contributing to this loss is the fact that other copular constructions, which allow a non-referential element and even an empty subject in precopular position, existed from very early on (1200–1300). The first are constructions with an adjectival type of word such as bueno es que ‘it is good that’, verdad es que ‘it is true that’, and such which predicate an attribute of the subordinate clause. The latter are conditional constructions without a subject which are attested already in Latin (si es que ha pecado, castigadle ‘if it is that he has sinned, punish him’).
3. PREVIOUS DIACHRONIC STUDIES
ON ES QUE
CONSTRUCTIONS
The most thorough diachronic analysis made on es que constructions is the one by Dufter (2008). From the analysis of data from the CORDE (Corpus Diacro´nico del Espan˜ol), Dufter concludes that es que constructions evolve from structures such as esto es que ‘this is that’, la cosa es que ‘the thing is that’, and ası´ es que/como ‘that is how’, with anaphoric subjects, which are later omitted as they are recoverable from the previous context. That is, how we arrive at the modern uses of es que. Dufter’s conclusions on the chronology of inferential sentences seem unquestionable, since they follow from the statistical analysis of empirical data. However, the hypothesis he starts from
152
Magdalena Romera
is debatable to some extent. His study is based on the hypothesis that inferential sentences are only related to structures such as esto es que, la cosa es que, and ası´ es que/como, but not to other constructions which contain copula þ que þ a subordinate clause. He therefore limits his search to cases where the subject of the copula is an anaphoric pronoun or an anaphoric adverb (cf. Dufter, 2008). Other cases with different types of subjects (definite NP, indefinite NP, and pseudo-clauses introduced by lo que) are not observed. Moreover, Dufter makes a distinction between inferential sentences and what he calls ‘identificational sentences with zero subject’ and does not see a relationship between the two types of sentences. I am more inclined to believe that there is in fact a relationship between the different uses of es que. For many authors (Fuentes Rodrı´guez, 1997; Romera, 2004, among others), there are reasons to believe that es que and lo que pasa es que ‘what happens is that’ may be related to some extent. Namely, the fact that in almost every instance of es que in modern Spanish, the form lo que pasa es que can be used without altering the meaning substantially. This should be reason enough to extend the search to all possible subjects.
4. EMPIRICAL STUDY 4.1. The data The data of my analysis come from El corpus del espan˜ol, a 100-million-word corpus of Spanish texts from 1200 to 1999, compiled by Mark Davis and accessible through the web (www.corpusdelespanol.org). The corpus contains 20 million words from the 1200s to 1400s, 40 million words from the 1500s to 1700s, 40 million words from the 1800s to 1900s, and 20 million words from the 1900s. The data analyzed in this paper belong to the period from 1200 to 1900. The searches made in the Corpus del Espan˜ol tried to find all possible structures in which es que appears as a sequence. Since an es que search by itself would have returned thousands of cases, partial searches were made by grammatical categories as shown below: VERB þ es que; NOUN þ es que; ADJECTIVE þ es que; ADVERB þ es que; PRONOUN þ es que; QUE þ es que; NO þ es que; SI ‘if’ þ es que; Y ‘and’ þ es que. Also, a last search was made to compile other possible cases in which es que was preceded by any other element: *es que. This last search returned cases in which es que followed a punctuation mark as well: point, coma, colon, and semicolon.5
5 For the purpose of this analysis, cases in which other discourse functional units preceded ES QUE were not looked at. Also, examples with other tenses of the copula were as well left aside.
Origin of es que
153
The results returned a total of 6330 valid cases, which were coded and analyzed using the statistical program SPSS (Statistical Package for the Social Sciences, v. 13). Each instance of es que was coded according to the following variables: 4.1.1. Formal variables. Form (i.e., the actual form itself, e.g., fuerza es que, ora es que, si es que, etc.); Grammatical category preceding es que; Ordering number in the database; Source; Date. 4.1.2. Functional variables. Type of structure: integrated/nonintegrated; Semantic category of verb preceding es que: thinking, saying, feeling, action, etc.; Subject: present/absent; Type of subject: noun, pronoun, clause; Degree of referentiality of subject; Relation of the es que construction with a previous discourse segment: definitional, elaboration, reinterpretation. 4.2. Results 4.2.1. Loss of referentiality. The data indicate that the subjects of the es que constructions undergo a loss of salience or referentiality. The es que structure focalizes the second argument of the copula (the subordinate clause), which results in the weakening of the first argument, where early on there is an increasing loss of referentiality. In my analysis, I adopt a model of reference based on three properties of linguistic forms: animacy, definiteness, and specificity (Von Heusinger and Kaiser, 2003). Animacy: A lexical feature of the linguistic expressions that expresses a property of the referent. The animacy scale would include humanWanimateWinanimate features. Definiteness: Definiteness is a discourse-pragmatic property that indicates that the discourse referent associated with a definite expression can be identified with an already introduced discourse item. It is a property of the referents set in the discourse. A definite singular expression unambiguously denotes or refers to one object, that is, the object can be identified as the only one that is denoted by the expression (Von Heusinger and Kaiser, 2003: 44). Specificity: The concept of specificity refers mainly to the more or less constrained reading of a referent. ‘‘Specific readings arise when the choice of referent is presented as heavily constrained, and nonspecific readings arise when the choice is presented as being relatively free’’ (Leonetti, 2003: 69). Within this model, I assume a slightly modified version of the scale of referentiality proposed in Aissen (2000) (Figure 8.1). As mentioned above, it seems that in es que constructions, the information load is moving from the subject position to the predicate position. The subject position goes from being medium informative to low informative. As a consequence, the elements placed in the subject position must be of weak referentiality, since they are in a non-prominent position. This process allows the subject position to be left empty.
154
Magdalena Romera
Anaphoric Pronouns > (specific, definite and with an accessible referent in the context; e.g. esto/esso/ello) Specific Definite NPs > (e.g. el quinto conssejo es que) Specific Indefinite NPs > (e.g. amor es que) Non-Specific Indefinite NPs > (e.g. cosa, caso) Pseudo-cleft clause (non-specific, indefinite, only accessible through the information following the copula; e.g. lo que hace al caso es que).
Figure 8.1. Scale of referentiality. Es que structures also undergo a process of subjectification (Langacker, 1990; Traugott, 1995), that is, the introduction of the speaker’s point of view in the expression of propositional content. So, although initially these constructions elaborate the content of a previous discourse segment, later on they are used to introduce a reinterpretation of what has been said in terms of the speaker’s opinion. Regarding the degree of referentiality of the element in the subject position, the results first indicate that the grammatical category appearing most frequently in all periods is the NOUN6 (Table 8.1). In Early Spanish (1200–1300) nouns are 89.4% and 83.7% of all the grammatical elements found, which means that in this period the referentiality of the subject is at the top of the scale. In contrast, this percentage decreases in following periods in favor of other types of subjects, more precisely, the pseudo-cleft clauses and the absence of a subject. This is also confirmed from a qualitative point of view; in the period 1200–1300, only the most referential nouns are used in this position: definite NPs, nouns specified by an adjective or a relative clause, or both: (12)
1300 Adjective þ Noun þ Relative clause: el sesto conssejo que el dicho Sennor alos medicos da [es que] . . . ‘the sixth piece of advice that the aforementioned Gentleman gives to the doctors [is that] . . . ’7 Visita y consejo de medicos. (Este´fano de Sevilla.)
As shown in Table 8.2, nonspecific nouns in subject position (e.g., tiempo es que, hora es que, razo´n es que, fuerza es que) become frequent from 1500 onwards. It is also the case of the nouns
6 Due to the different extension of texts in each period, percentages are presented within the same period. This allows for comparison of frequencies of use between periods. 7 Glosses are not meant to be literal but approximate translations, as many of the words are not used in the same sense in Modern Spanish or have disappeared altogether from the language.
Origin of es que
155
Table 8.1. Percentage of subject types within each diachronic period Subject type
PERIOD
TOTAL
1200
1300
1400
1500
1600
1700
1800
281
149
335
676
344
249
614
NOUN N %
89.4
83.7
87.5
77.6
16
24
74
59.5
45.2
13
46
2648
ANAPHORIC PRONOUN N
31
22
%
10
12.3
4.1
2.7
8 1.7
3.1
160
3.3
PSEUDO-CLEFT CLAUSE N
2
7
28
155
93
110
%
0.6
4
7.4
17.9
20
26.4
0
0
4
16
20
46
351
746
25.8
NO SUBJECT N %
1
1.8
4.3
11
348
434
25.7
TOTAL N
314
178
383
871
465
418
1359
%
100
100
100
100
100
100
100
3988a
a
This table shows results for those cases in which the referential status of the elements in subject position could be measured. Cases corresponding to non-referential copular constructions and conditional structures were not computed.
caso ‘case’ and cosa ‘thing’. Both can be considered of a low referentiality, since at least in theory they are inanimate, nonspecific, and indefinite most of the time. However, in Early Spanish, both nouns appear specified by an adjective or have meaning of occasion or situation (e.g., el quinto caso es que ‘the fifth case is that’). The indefinite, nonspecific use of caso as a subject of an inferential (meaning ‘the thing is’) construction is observed from 1500 onwards, becoming its most frequent use. The same can be said of cosa. In 1300, we find cases of cosa specified by an adjective: manifiesta/sabida cosa es que ‘manifested/known thing is that’, but uses of cosa without specification appear since 1600.
156
Magdalena Romera
Table 8.2. Semantic types of NPs in subject position in each period Semantic type of NP
PERIOD 1200
1300
1400
1500
152
299
TOTAL 1600
1700
1800
108
282
SPECIFIC DEF NPS N %
177 59.8
75 43.1
47.6
55.6
56 26.1
32.5
36.5
61
48
74
28.3
14.5
1149
SPECIFIC INDEF NPS N %
103
59
135
34.8
33.9
42.3
16
40
32
23
10.1
136 25.3
616
9.5
Non-SPECIFIC IND N %
5.4
103 19.1
98 45.6
176
418
53
54
883
TOTAL N
296
174
319
538
215
332
774
%
100
100
100
100
100
100
100
2648
Therefore, it seems that the nouns appearing in subject position of es que constructions lose their referentiality over time. We also find that the subject position can be occupied by an anaphoric neuter pronoun. These pronouns are highly referential, since their referent is immediately accessible in the previous context and there is no need of additional information to unmask their meaning. From 1200 to 1800, the most frequently used neuter pronouns are ESTO ‘this’ and ELLO ‘that’. ESO ‘that’ competes with ESTO from 1800 onwards. However, regarding their frequency, we observe that their use decreases over time with respect to other subject types: from 12.3% in the period 1200– 1300 to 3.3% in 1800. This again supports our hypothesis that the elements in the subject position lower their referential status. Also, their overall low frequency with respect to other grammatical subject types indicate that there are in fact more prominent structures adding to the formation of inferential sentences with es que. Finally, the results indicate a progressively higher frequency of pseudo-cleft clauses. In these cases, only the predicate position (after the copula) can convey salient information. The subject
Origin of es que
157
position seems to function as an introductory part of the truly relevant information to come. In contrast to previous cases, the referents of these elements are opaque, undetermined, and nonspecific. To some extent they share properties with cataphoric expressions, since their referent is only made explicit through the information given in the following discourse. It is in this sense that we can talk of a loss of referentiality in the subject position. Pseudo-cleft clauses are very infrequent in Early Spanish: only 2 cases out of 6,905,000 words, but as Table 8.1 shows, their frequency increases considerably over time to the point of being the second most frequent subject type after nouns in 1800 and almost equal in percentage to es que constructions with no subject. A factor that may have contributed to the reinforcement of the loss of the subject in es que structures is the existence of non-referential copular constructions and conditional es que expressions without a subject. From very early in the history of Spanish, non-referential structures with the copula are frequently used. Examples of menester es que ‘it is necessary that’, cierto es que ‘it is true that’, and posible es que ‘it is possible that’ are found in the corpus from 1200 to 1300 onwards as shown in Figure 8.2. Even closer in structure to the present day es que construction is the expression formed by the conditional complementizer si ‘if’ followed by a copula and que, as in (13): (13)
Si es que ha pecado, castigadle ‘If it is that he has sinned, punish him’
Si es que structures are traced already in Latin as in (14) (Jorda´n Co´lera, p.c.): (14)
Si est ut uelit redducere uxorem, licet ‘If it is that he wants to take again his wife, I agree’ Terence, The Mother-in-Law
100 Non-REF CONSTR SI CONSTR Non-INTEGR ES QUE
80 60 40 20 0 1200
1300
1400
1500
1600
1700
1800
Figure 8.2. Percentage of non-referential and conditional constructions found in corpus.
158
Magdalena Romera
Although not very frequent in our corpus, this structure appears in Early Spanish with some orthographic variation. The following example shows one of the cases found: (15)
le puede vedar que dexe de labrar en aquella labor fueras ende si [es que] gelo vedase fuese huerfano ‘he can order him to stop working the field in that labour outside the field if [it is that] he prohibited it to him, he was an orphan’ (Alfonso X, Siete partidas)
From 1500 onwards, the structure is more frequent and is used as in Modern Spanish. The existence of such structures without a subject for the copula could contribute to the use of subjectless es que constructions. This is confirmed by the appearance of cases in which es que follows a punctuation mark, that is, lacking a subject, in the same period.8 The first cases of nonintegrated constructions are found in 1500 in monologues and at the end of that century in dialogues. (16)
La infanta, que luego oyera rencilla tan grande e brava, a una de las damas suyas lo que era preguntaba. [Es que] rey, sen˜ora, al conde de traidor lo difamaba ‘The princess, that later heard such a big and strong quarrel, to one of her ladies what that was she asked. [It is that] the king, My Lady, accused the count of being a traitor.’ (Anonymous. Rom. Viejo.)
A search was also made for cases of y þ es que, clearly used in focalizing structures, where the element in predicate position (after es que) is emphasized as it presents the relevant information. This structure also appears in 1500. (17)
La.iij. sen˜al es mas de proximo que ninguna otra segund algu nos. y [es que] quando de noche suen˜an los lapidosos que comen . . . . ‘The 3rd sign is closer than any other according to some. And [it is that] when at night the sick people dream that they eat . . . ’ (Julia´n Gutie´rrez de Toledo. Cura de la piedra.)
Therefore, it seems that the subject of es que constructions indeed undergoes a progressive loss of referentiality from Early Spanish to the 16th century, where it is possible to leave this position empty. Other constructions including a conjunction ( y ‘and’) followed by es que also appear without a subject. The identity in terms of relevance between the information in subject and predicate positions is lost progressively as well. The information load is placed on the predicate position, and the subject position is only used to formally introduce the upcoming content. This in 8 In modern examples of es que (i.e., without a subject), this expression often appears as introducing a speech turn or in a monologue following a punctuation mark (mostly a comma or a point).
Origin of es que
159
turn allows for the empty subject position. The process of loss of referentiality argues against the theory that the referents of subjects of es que constructions are easily recoverable (Declerck, 1984, 1992; Ferna´ndez Leborans, 1992). In fact, it seems that there is no referent to recover from a previous context; rather, the element placed in the subject position is more a formal formula to introduce the next member. Its low value allows its omission. 4.2.2. Subjectification. Hand in hand with the process of loss of referentiality, there is a process of subjectification of the relation established between the es que clause and the previous discourse unit. Es que constructions are in origin definitional and elaborative structures, from Early Spanish up to the 16th century. The definitional value describes a fact or defines a term. The subject is explicit and its definition appears in the predicate position. The structure does not add any other meaning besides the presentation of what a subject is. Apart from their definitional use, es que constructions also introduce an elaboration of the content expressed in the previous utterance(s). Moreover, this elaboration can be made in terms of the speaker’s point of view, that is, a previous statement can be reinterpreted according to the speaker’s opinion, as an explanation of what has just been said. Table 8.3 shows percentages of occurrence of the three functions of es que constructions: definitional, elaboration specific, and reinterpretation. The following is an example of a definitional use of es que from 1300. (18)
y le dixo la voluntad de dios [es que] tu passes los montes de ballem ‘and he told him the will of God [is that] you climb past the mountains of ballem’ (Juan de Mandevilla. Libro de las maravillas del mundo.)
The elaboration cases can introduce a specification of a list of terms, all belonging to a previous generic or global term. (19)
a en si quatro uertudes. /[ La primera es que ]al que la troxiese consigo; nunqual fallaron muerto en agua. ‘It has in it four virtues./ [the first is that] whoever took it with him; he was never found dead in the water.’ (Alfonso X. Lapidario.)
Especially clear is the elaboration specific use of es que constructions when anaphoric pronouns are subjects. (20)
primera de las siete dificultades se funda sobre que la vida es breue./[Esto es que] la vida del medico es breue para poder alcanc- ar las perfecc- iones que le conujene ‘First of the seven difficulties is based on (the fact) that life is short. [This is that] a doctor’s life is short so as to be able to reach the perfections that suit him’ (Alfonso Chirino. Espejo de medicina. 1400)
160
Magdalena Romera
Table 8.3. Percentage of functional values of es que constructions within each period Functional value
PERIOD 1200
1300
1400
TOTAL
1500
1600
1700
1800
95
23
16
26
DEFINITIONAL N
178
%
97
49.4
41.5
156 26.1
7.6
2.8
2.3
591
1.3
ELAB SPECIFIC N %
177
94
49.2
298
40.2
49.9
28
44
434 34.5
111 13.3
162 23.6
156
1432
7.6
REINTERPRETATION N
0
%
12
7.4
498 39.6
596 71.3
374 54.4
1620
3160
78.9
DEFINIT/ELAB/REINTERP N
5
%
1.4
15 6.4
99 16.5
231 18.1
109 12.7
135 18.7
252
843
12.2
Total N
360
234
597
1258
836
687
2054
%
100
100
100
100
100
100
100
6026a
a Conditional constructions with si þ es que were excluded from this analysis since they played a different function in discourse.
The first cases of nonintegrated es que structures (without a subject) are used to convey the elaboration-specific meaning, which can be presented as a clarification of the previous term. (21)
una cosa no han considerado los filo´sofos naturales [ . . . ] [ Y es que ] las cosas que tienen mucha sal . . . ‘One thing have not considered natural philosophers [ . . . ] [And that is] that things that have a lot of salt . . . ’ (Huarte de San Juan, Juan. Examen de ingenios para las ciencias, 1558)
Origin of es que
161
From this elaboration-specific meaning, a more subjective use develops. When the speaker introduces the elaboration clarification segment, it may be interpreted in terms of her own beliefs or opinions. The explanation glosses by Pe´rez de Moya in the 16th century below are clear examples of this reinterpretation. (22)
(23)
por esto dijeron que estos a´rboles sudaron a´mbar. Esto, segu´n histo´rico sentido, [es que] como Phaeto´n muriese en aquel viaje . . . ‘for this they said that these trees sweated amber. This, in historic sense, [is that] since Phaeton died in that trip . . . ’ el Sol, que es Apolo, al signo de Sagitario, casa de su padre, pone lechugas a Iuno, [ es que como ] las lechugas son frı´as, ası´ en este tiempo la tierra, entendida agora por Iuno, se ocupa con frı´os ‘the Sun, which is Apollo, to the Saggitarius sign, home of his father, puts (offers) lettuces to Juno, [it is that] since lettuces are cold, thus in this time the earth, known now by Juno, is occupied by (full of) cold weather.’ (Pe´rez de Moya, Juan. Philosofı´a secreta, 1554)
From the 16th century, the subject position is progressively less referential and is already left empty. Note that example (22) presents an anaphoric pronoun, but (23) does not have an explicit subject anymore. In the 16th century, the reinterpretation segments are still of an objective nature, where the explanation is close in meaning to the term/content stated previously, as in (24). (24)
Y si nos cargan, [es que] cuando van de camino les llevamos su hatillo y algunos libros para predicarnos, que todo ello no pesa nada. ‘And if they load us, [it is that] when they go on a trip we carry their bags and some books to predicate us, that all that is not heavy at all.’ (Mendieta, Jero´nimo de, Historia eclesia´stica indiana, 1564)
Later on, the reinterpretation segments progressively acquire a higher subjective character as in example (25), from 1700, where it is clear that the es que construction reinterprets the unit preceding es que in terms of the speaker’s belief of what must have happened. (25)
Sintio´se mucho de esto don˜a Isabel, y [es que] debio´ entender, que por ser hermano suyo no habı´a en la disciplina militar para e´l cosa limitada. ‘Miss Isabel was very sorry about this, and [it is that] she must have understood, that because he was her brother there was not in the military discipline for him a limited thing.’ (Queiros, Pedro Fernandes de., Historia del descubrimiento de las regiones austriales, 1590)
It is well known that the acquisition of new meanings does not imply the immediate disappearance of the original meaning of an expression. Rather, old and new meanings tend to coexist for long periods of time, sometimes lasting for centuries. In our data this is confirmed by
162
Magdalena Romera
cases in which various meanings can be attributed to es que constructions at the same time. Es que appears in bridging contexts (Evans and Wilkins, 1998; Diewald, 1999; Heine, 2002), in which the new meanings, elaboration, or reinterpretation are already inferable, but the original definitional meaning cannot be ruled out. As can be observed in Table 8.3 (above), these cases are represented in all periods. The progressive subjectification of es que constructions is also supported by other signs of subjectivity. From the 16th century onwards, nouns in the subject position explicitly express the introduction of the speaker’s opinion with formulas such as El mı´o parecer es que ‘my opinion is that’ and la opinio´n general es que ‘the general opinion is that’. The same can be said of pseudo-cleft clauses, which become frequent from the 15th century onwards. They are used to introduce the speaker’s beliefs or knowledge. Thus, the majority of verbs used in these clauses belong to the semantic category of cognition (lo que sabemos es que ‘what we know is that’, lo primero que habe´is de presuponer es que ‘the first thing you have to presuppose is that’, lo que yo veo es que ‘what I see is that’). Our data show that pseudo-cleft clauses used in es que constructions in all periods present 71.9% of verbs of a subjective nature (26.2% are verbs of thinking or opinion, 23.4% verbs of saying, and 22.3% verbs of feeling) versus 17.8% of verbs of movement/action and 10.3% of verbs of state. Finally, subjective meanings can also be introduced when the contents of the units related are interpreted epistemically (Sweetser, 1990). That is, the expression of the cause of a certain event or a real-world situation is interpreted as the reason why something has been said. In example (25) above, the cause of Mrs. Isabel’s sorrow can be viewed as part of the speaker’s reasoning on how the events developed. In Sweetser’s words, ‘‘as descriptions, sentences describe real-word events and the causal forces leading up to those events; as conclusions, they are themselves understood as being the result of the epistemic forces which cause the train of reasoning leading to a conclusion’’ (Sweetser, 1990: 65).
5. CONCLUSIONS To sum up, my proposal that es que constructions endure a process of subjectification and a loss of referentiality of their subjects is supported by the data. We have seen that es que structures are initially definitional and elaborative. Then, a more subjective interpretation in terms of the speaker’s beliefs is introduced, which is understood as an explanation for what has just been said. At the same time, the subjects of es que constructions gradually lose their referential status. Initially, they are highly referential and can be related anaphorically to the previous content, but later they are simply elements that anticipate the content of the subordinate clause.
Origin of es que
163
Chronologically, these two processes go together. From Early Spanish (1200–1300) up to the 16th century, es que constructions are definitional and elaborative structures. The first cases of interpretative uses are found in 1500 and become common in the 17th century. The same can be said of the path from integrated constructions (i.e., es que constructions with a subject) to the nonintegrated ones (i.e., es que constructions without a subject). No cases of subjectless constructions were found in Early Spanish and only a few cases in 1400. The first uses of es que structures as we know them in Modern Spanish appear in 1500 in monologues and at the end of that century in dialogues. The subject position is left empty in the 16th century.
ACKNOWLEDGMENTS I am grateful to Andreas Dufter, Gorka Elordieta, Lucı´a Loureiro, Claus Pusch, Cristina Sua´rez, the late Andreas Wesch, and an anonymous reviewer for helpful comments. Special thanks must go to Maj-Britt Mosegaard Hansen for her generous help. Partial funding for this work came from the financial assistance granted by the Spanish Ministerio de Ciencia e Innovacio´n (HUM2006-12695).
REFERENCES Aissen, J. (2000). Differential object marking: Iconicity vs. economy, Ms, UCSC. Declerck, R. (1984). ‘‘The pragmatics of it-clefts and WH-clefts’’. Lingua 64: 433–471. Declerck, R. (1992). ‘‘The inferential it is that-construction and its congeners’’. Lingua 87: 203–230. Delahunty, G. P. (2001). ‘‘Discourse functions of inferential sentences’’. Linguistics 39: 517–545. Delahunty, G. P. and L. Gatzkiewicz (2000). ‘‘On the Spanish inferential sentences’’. Pragmatics 10 (3): 301–322. Diewald, G. (1999). Die Modalverben im Deutschen: Grammatikalisierung und Polyfunktionalita¨t. Tu¨bingen: Niemeyer. Dufter, A. (2008). ‘‘Evolucio´n pragma´tica de las oraciones hendidas en espan˜ol. El papel de los usos no focalizadores’’, Actas del VII Congreso Internacional de Historia de la Lengua, Me´rida 2006, 1763–1780. Espan˜a Villasante, M. (1996). ‘‘Aspectos sema´ntico-pragma´ticos de la construccio´n ‘es que’ en espan˜ol’’. Dicenda 14: 129–147. Evans, N. and Wilkins, D. (1998). The knowing ear: An Australian test of universal claims about the semantic structure of sensory verbs and their extension into the domain of cognition (Arbeitspapiere von Institut fu¨r Sprachwissenschaft Universita¨t zu Ko¨ln, Neue Folge 32). Ko¨ln: Institut fu¨r Sprachwissenschaft. Ferna´ndez Leborans, M. J. (1992). ‘‘La oracio´n del tipo: ‘es que . . . ’’’. Verba 19: 223–239. Fuentes Rodrı´guez, C. (1997). ‘‘Los conectores en la lengua oral: es que como introductor de enunciado’’. Verba 24: 237–263.
164
Magdalena Romera
Hedberg, N. (1990). Discourse pragmatics and cleft sentences in English. Ph.D. dissertation, University of Minnesota. Heine, B. (2002). ‘‘On the role of context in grammaticalization’’, in I. Wischer and G. Diewald (eds.), New reflections on grammaticalization. Amsterdam: John Benjamins, 83–101. Hobbs, J. (1990). Literature and cognition, lecture notes 21. Stanford, CA: Center for the Study of Language and Information. Hobbs, J. (1996). ‘‘On the relation between the informational and intentional perspectives on discourse’’, in E. Hovy and D. Scott (eds.), Computational and conversational discourse: Burning issues, an interdisciplinary account. Berlin: Springer Verlag, 139–157. Hovy, E. and D. Scott (eds.) (1996). Computational and conversational discourse: Burning issues in discourse (NATO series F, no. 151). Heidelberg: Springer Verlag. Langacker, R. W. (1990). ‘‘Subjectification’’. Cognitive Linguistics 1: 5–38. Leonetti, M. (2003). ‘‘Specificity and object marking: The case of Spanish a’’, in K. von Heusinger and G. A. Kaiser (eds.), Proceedings of the Workshop Semantic and Syntactic Aspects of Specificity in Romance Languages. Konstanz: University of Konstanz, 67–102. Mann, W. C. and S. Thompson (1988). ‘‘Rhetorical structure theory: Toward a functional theory of text organization’’. Text 8: 243–281. Marcu, D., Romera, M. and Amorrortu, E. (1999). ‘‘Experiments in constructing a corpus of discourse trees: Problems, annotation choices, issues’’, in Proceedings of the Workshop on Levels of Representation in Discourse. Edinburgh: University of Edinburgh, 71–78. Partee, B. H. (1986). ‘‘Ambiguous pseudoclefts with unambiguous be’’, in S. Berman, J. Choe and J. McDonough, J. (eds.), Proceedings of NELS 16, GLSA. Amherst, MA: University of Massachusetts, 354–366. Polanyi, L. (1996). The linguistic structure of discourse, Technical Report CSLI-96-200. Stanford, CA: Center for the Study of Language and Information. Romera, M. (2004). Discourse functional units. Munich: Lincom Europa. Sanders, T., W. Spooren and L. Noordman (1992). ‘‘Towards a taxonomy of coherence relations’’. Discourse Processes 15: 1–35. Sweetser, E. (1990). From etymology to pragmatics. Metaphorical and cultural aspects of semantic structure. Cambridge: Cambridge University Press. Traugott, E. C. (1995). ‘‘Subjectification in grammaticalization’’, in D. Stein and S. Wright (eds.), Subjectivity and subjectivisation in language. . Cambridge: Cambridge University Press. Von Heusinger, K. and Kaiser, G. A. (2003). ‘‘The interaction of animacy, definiteness, and specificity in Spanish’’, in K. von Heusinger and G. A. Kaiser (eds.), Proceedings of the Workshop Semantic and Syntactic Aspects of Specificity in Romance Languages. Konstanz: University of Konstanz, 41–66.
Current Trends in Diachronic Semantics and Pragmatics Edited by Maj-Britt Mosegaard Hansen and Jacqueline Visconti r 2009 by Emerald Group Publishing Limited. All rights reserved.
9 FROM ASPECT/MOOD MARKER TO DISCOURSE PARTICLE: RECONSTRUCTING SYNTACTIC AND SEMANTIC CHANGE Bethwyn Evans
1. INTRODUCTION This paper examines the reanalysis of an aspect/mood marker as a discourse connective particle from the perspectives of both syntactic and semantic change. Evidence of the change is found in the system of subject marking in Marovo, an Oceanic language of the Solomon Islands. Marovo has preverbal markers which indicate the person and number of the subject argument and occur primarily in only two types of constructions: negative verbal declarative clauses and verbal clauses with an initial discourse connective particle. These unusual conditions on the presence of subject marking in Marovo are shown to reflect its historical development. Through comparison of Marovo with other closely related Oceanic languages, it is demonstrated that subject marking in negative clauses is archaic, reflecting original constructions in which subject markers occurred within the verb complex alongside preverbal markers of aspect/ mood and negation. The use of subject markers with discourse connective particles reflects the same original construction, but in this case the reanalysis of an aspect/mood marker as a discourse connective particle has resulted in the subsequent extension of subject markers to use with discourse connective particles in general. A detailed reconstruction of the change, informed by accepted models of syntactic and semantic reanalysis (Harris and Campbell, 1995; Traugott and Dasher, 2002; Eckardt, 2006), suggests that it was motivated by both the morphosyntactic and semantic characteristics of the
166
Bethwyn Evans
construction. While the reanalysis appears to have been triggered by structural ambiguity resulting from a chance homophony of forms, semantic and pragmatic aspects of the construction also facilitated the change. The details of this particular change are presented in Section 5, where data from Marovo and the closely related language Roviana are used to reconstruct the syntactic and semantic contexts, which motivated the change. This section also demonstrates the ways in which the present reconstruction has been informed by established models of language change. Preceding sections provide the synchronic and diachronic context within which the change occurred. Thus Section 2 provides an overview of Marovo and its place within the Oceanic language family, and Section 3 describes the use and distribution of subject markers in Marovo, giving details of the construction that represents the end result of the reanalysis. Section 4 gives an overview of the development of subject marking in Marovo highlighting its archaic and innovative characteristics with respect to Proto Oceanic.
2. MAROVO
AND THE
NEW GEORGIA LANGUAGES
Marovo is spoken on islands in and around Marovo Lagoon in the Western Province of the Solomon Islands. It is a member of the Oceanic subgroup of languages within the large Austronesian family. Within the Oceanic subgroup, Marovo is part of the Northwest Solomonic linkage that also includes languages of the New Georgia, Shortland and Choiseul islands in the Solomon Islands and of Bougainville, Buka and Nissan islands of Papua New Guinea.1 Within this linkage Marovo is most closely related to other languages of the New Georgia group of islands, and in particular those of the eastern region. The linguistic history of Marovo and the development of particular constructions are presented in terms of retention and innovation from several reconstructed proto-languages, namely Proto Oceanic, Proto Northwest Solomonic, Proto Eastern New Georgia and pre-Marovo. Proto Eastern New Georgia is the hypothesised common ancestor of the Oceanic languages in the eastern region of the New Georgia group of islands, including Roviana, Ughele, Hoava, Kusaghe, Marovo, Vangunu and Bareke. While I am not claiming that the evidence for the existence of a distinct Proto Eastern New Georgia speech community is completely conclusive (but see Ross, in press), such a stage is used to represent the shared history of these eastern languages. The system of
1 Ross (1988: 8) uses the term linkage to refer to a group of languages which have arisen through dialect differentiation, such that the set of innovations which define the group are not necessarily shared by all languages of the group. Rather the innovations have different, but overlapping, geographical domains. This is in contrast to subgroups that are defined by innovations which are shared by all members of the group.
From aspect/mood marker to discourse particle
167
Table 9.1. Languages and proto-languages relevant to the present study Proto Oceanic Proto Northwest Solomonic Proto New Georgia Proto Eastern New Georgia
Ancestor of all Oceanic languages Ancestor of languages of Bougainville and western Solomon Islands Ancestor of languages of New Georgia group of islands Ancestor of languages of eastern New Georgia region, incl. Roviana, Hoava and Marovo
aspect/mood and subject marking in western New Georgia languages is rather different from that in the eastern languages, and further investigation is needed to reconstruct accurately the Proto New Georgia system. Pre-Marovo represents structures reconstructed within the history of Marovo, but distinct from other eastern New Georgia languages. The reconstruction of preMarovo constructions with subject marking may reflect a shared history with Vangunu (see Bourchier, 2007), but again further investigation is needed. The linguistic relationships and protolanguages relevant to this study are set out in Table 9.1. The New Georgia languages considered in detail in the present study are Roviana, Hoava and Marovo. Detailed descriptions are available for Hoava (Davis, 2003) and Roviana (Waterhouse, 1949; Corston, 1996; Corston-Oliver, 2002, 2003), and all Marovo data presented are based on my own fieldwork. Certain changes, relevant to the present study, have occurred in Roviana between the time of Waterhouse’s and Corston-Oliver’s descriptions of the language, and so a distinction is made between Early Roviana (data from Waterhouse, 1926, 1949; Western Province Government, 1991) and (contemporary) Roviana (data from Corston, 1996; Corston-Oliver, 2002, 2003). The basic clause order in Marovo, and indeed other New Georgia languages, is VERB-SUBJECT2 OBJECT, as demonstrated by (1). Marovo [hami]SUBJ (1) [Heru-i]VC 1PLEXCL carry-TR:3PLO ‘We carried the timber . . . ’
[ria ART:PL
labete]OBJ . . . timber
The syntactic string which comprises the lexical verb or verbs, any accompanying adverbiallike modifiers and morphemes marking aspect, mood, transitivity and participant reference will be labelled the verb complex. For ease of interpretation the verb complex in the examples presented
2
Abbreviations follow the Leipzig Glossing Rules, and additionally include: ASP – aspect marker, DES – optative mood, INTJ – interjection, O – object marker, P – possessive pronominal form, PC – paucal marker, RDP – reduplication, RL – realis mood, S – subject marker, VC – verb complex.
168
Bethwyn Evans
here is enclosed in square brackets and labelled VC. The subject and object noun phrases are also enclosed in square brackets and labelled SUBJ and OBJ respectively.
3. SUBJECT MARKERS
IN
MAROVO3
Marovo, like many other Oceanic languages, has preverbal markers which indicate the person and number of the subject argument. Thus in the intransitive clause in (2) the form gu indicates that the subject argument is a 1st person singular (1SG) participant; this information is also expressed by the clause-final independent pronoun raka. (3) is a transitive clause and demonstrates that the preverbal subject markers may also index transitive subjects. Marovo (2)
Ma ¼ [gu jama]VC then ¼ 1SGS speak ‘Then I said . . . ’
(3)
Ma ¼ [gu then ¼ 1SGS ‘Then I called Junior.’
la ASP
[raka]SUBJ. 1SGS
ole-a]VC call-3SGO
[raka]SUBJ 1SGS
[Junior]OBJ J.
These subject markers can co-occur with a pronominal or lexical noun phrase, or may be the only expression of the subject argument within the clause. Examples (2) and (3) illustrate the use of subject markers with pronominal noun phrases. Example (4) shows the use of the 3SG subject marker ni with a lexical noun that expresses the subject, while in (5) ni is the only expression of the subject argument within the clause. Marovo (4)
(5)
3
He ¼ [ni kala]VC go th/fore ¼ 3SGS ‘Therefore the turtle went.’
[vonu]SUBJ. turtle
Ma ¼ [ni choga va-kiki then ¼ 3SGS jump CAUS-small ‘Then he jumped gently into the sea . . . ’
la]VC go
pa LOC
idere sea
...
A more detailed account of the use and distribution of subject markers in Marovo is provided in Evans (2008).
From aspect/mood marker to discourse particle
169
Although, there are a few examples in which subject markers appear to be phonologically independent (see example (10b)), typically they form a phonological word with the preceding discourse connective particle, as indicated by the patterning of primary stress in (6) and (7).4 In (6) the connective particle pata ‘in order that’ occurs with primary stress on the penultimate syllable, the usual stress pattern in Marovo. In (7), pata ‘in order that’ occurs with a following subject marker and the sequence pata ¼ gu behaves as a single phonological word in terms of stress: primary stress occurs on the penultimate syllable of the entire sequence. Marovo (6)
. . . ma-[gu valu keli]VC pa´ta then-1SGS paddle go.up in.order ‘ . . . then I paddled up to go fishing . . . ’
(7) Pata´ ¼ [gu gura be.able in.order ¼ 1SGS ‘For me to be able to net.’
[la go
chaba]VC . . . to.fish
vagara]VC. to.net
Unlike preverbal subject markers in many Oceanic languages, in Marovo such markers are not obligatory, but have a rather restricted distribution. Within the set of narrative data examined here less than half the clauses included subject markers. In many clauses the expression of the subject argument is a pronominal or lexical noun phrase only, as in (8), or the subject argument is inferred from context and not overtly expressed within the clause at all, as in (9b). Marovo (8) [Taleto ni-a]VC [vonu]SUBJ [ia feel.sorry TR-3SGO turtle ART:SG ‘Turtle felt sorry for this wife . . . ’ (9) a. b. a. b.
...
beto [pocho]VC [ia]OBJ finish squeeze 3SG [va-reka la ni-a]VC pa CAUS-hot go TR-3sgO LOC ‘ . . . after we have squeezed it [the coconut],’ ‘we boil it in the fire.’
ororeke wife
pia]OBJ
...
DEM
[hami]SUBJ, 1PLEXCL ikuchu. fire
The presence of subject markers within a clause in Marovo appears to be motivated by both discourse functions and the semantic-grammatical structure of the clause. A primary function of 4
The term discourse connective particle refers to a small class of words in Marovo that function to connect stretches of discourse and that have meanings which denote the relationship (e.g. temporal, causal etc.) between the events described.
170
Bethwyn Evans
subject markers in Marovo is reference tracking. Thus subject markers typically denote a referent that is the continuing topic of a section of discourse.5 For example, in (10) the 1SG participant is initially denoted by an independent pronoun. This referent continues as topic of the subsequent clauses, where it is denoted solely by the subject markers. Marovo (10) a. [Mae]VC [raka]SUBJ, come 1SG b. [gu mae kaduvu]VC 1SGS come arrive c. beto ma ¼ [gu la finish then ¼ 1SGS go d. ma ¼ [gu la mae come then ¼ 1SGS go a. ‘I came’ b. ‘and reached Adado Point,’ c. ‘and then I came over,’ d. ‘then I came and arrived at my shore.’
pa LOC
ukala past hodoko]VC arrive
chopochopo Adado, point A. mae]VC, come pa gua sera. LOC 1SGP shore
The presence of subject markers in Marovo is also conditioned by the semantic-grammatical structure of the clause in that the use of certain particles within the clause require the use of subject markers. Thus subject markers obligatorily occur with the negative particle ka-, as shown in (11). Marovo (11)
‘‘Oh,
[pavu]VC [ia]SUBJ, boru [ka ¼ ni tavete]VC’’, be.sick 3SG and.so NEG ¼ 3SGS work ‘‘‘Oh, he’s sick and so he’s not working’’, she said.’ INTJ
[hua]VC say
[ia]SUBJ. 3SG
Subject markers also always occur with certain discourse connective particles, namely ma ¼ ‘and then’, as seen in (2) and (3), and with he ¼ ‘therefore’, as in (12). With boru ‘and so’ and pata ‘in order that’, on the other hand, subject markers are optional (see examples (6) and (7)). Marovo (12)
Boru he ¼ [gu ka-gu vagara]VC and.so th/fore ¼ 1SGS NEG-1SGS to.net ‘So therefore I didn’t net at Omo . . . ’
pa LOC
Omo O.
[raka]SUBJ 1SG
...
5 The topic of a clause is defined as having three characteristics: (i) it is what the statement is about; (ii) it invokes ‘‘knowledge in the possession of an audience’’; and (iii) ‘‘the statement is assessed as putative information about its topic’’ (Strawson, 1964: 97–98; Erteschik-Shir, 2007: 13).
From aspect/mood marker to discourse particle
4. ORIGIN
AND
DEVELOPMENT
OF
SUBJECT MARKERS
IN
171
MAROVO
The distribution of subject marking in Marovo appears unusual when compared with that in other Oceanic languages. However, it can be shown to reflect the original Proto Oceanic system, and its unusual features explained by a series of natural diachronic developments. Preverbal subject markers similar to those in Marovo are found in a wide range of contemporary Oceanic languages, and can be reconstructed for Proto Oceanic. In Proto Oceanic, subject proclitics occurred as the first or second element within the verb complex, as shown by the structure of the Proto Oceanic verb complex reconstructed by Lynch et al. (2002: 83) in Table 9.2 This original structure is still clearly reflected in many Oceanic languages, including Hoava. In Hoava the preverbal future tense marker ma- can occur with 1SG or 1PLINCL subject markers, as in (13). Hoava (13)
Koleo, [ma-qu puta]VC. good FUT-1SGS sleep ‘Good, I will sleep.’ (Davis, 2003: 150)
Comparison of the distributions of cognate subject markers in Marovo, Hoava and Roviana, suggests that their common ancestor had a system of subject marking very similar to that reconstructed for Proto Oceanic. In both Hoava and Roviana the use of preverbal subject markers is marginal, and in fact is present in Early Roviana only.6 As mentioned, 1SG and 1PLINC subject markers occur with the preverbal future tense marker ma- in Hoava, and subject markers of all person/number categories occur with the preverbal optative marker o- (Davis, 2003: 149–152), as in (14).
Table 9.2. The Proto Oceanic verb complex (Lynch et al., 2002: 83) (ASPECT/MOOD ¼ )a subject marker ¼
VERB ( ¼ OBJECT
MARKER)
( ¼ DIRECTIONAL
MARKER)
a
The initial element of the Proto Oceanic verb complex, labelled as ASPECT/MOOD, does not include tense as it is presumed that Proto Oceanic, like many contemporary Oceanic languages, had grammatical distinctions for aspect and mood, but not tense (Lynch et al., 2002: 84)
6
While Waterhouse (1926, 1949) describes the use of subject markers in Roviana, more recent work (Corston, 1996; Corston-Oliver, 2002, 2003) makes no mention of preverbal subject markers; suggesting that the rather marginal system described by Waterhouse is no longer used.
172
Bethwyn Evans
Hoava (14)
[O-na napo]VC [sa OPT-3SGS drink ART:SG ‘The child wants to drink.’
koburu]SUBJ. child (Davis, 2003: 151)
Subject markers occur in cognate structures in Roviana, that is with the preverbal particles o- ‘desiderative’, as in (15) and ma- ‘imperative’, as in (16) (Waterhouse, 1949: 244–246). Roviana (15)
[Ma-mu IMP-2SG
podek-i-a]VC. try-TR-3SGO
‘You try it.’ (Waterhouse, 1949: 68)
(16)
[O-da
gani eat ‘We (incl.) wish to eat fish.’
DES-1PLINCL
igana]VC. fish (Waterhouse, 1949: 83)
In Roviana, ma- occurs with subject markers of all person/number categories. While with a 2nd person subject, ma- has an imperative function, with non-2nd person subjects it has the meaning of ‘to let, allow’, as in (17). Ma- in Roviana is also used to indicate future tense, as in (18). Roviana (17)
[Ma-qu IMP-1SGS ‘Let me see it.’
vilit-i-a]VC. see-TR-3SGO (Waterhouse, 1949: 245)
(18)
Uve; [ma-qu yes FUT-1SG ‘Yes, I will try.’
podek-i-a]VC. try-TR-3SGO (Waterhouse, 1926: 6)
Negative imperative constructions in Roviana also occur with preverbal subject marking. Negative imperatives with a 2nd person subject occur with the particle meke followed by the subject marker and then the verb, as in (19). In negative imperative constructions with a non-2nd person subject, the mood marker ma- plus a following subject marker are used alongside the usual negative particle lopu, as in (20).
From aspect/mood marker to discourse particle
173
Roviana (19)
[Meke NEG.IMP ‘Do not go.’
mu 2SG
la]VC. go (Waterhouse, 1949: 246)
(20) [Ma-qu lopu la]VC. IMP-1SGS NEG go ‘Let me not go.’ (Waterhouse, 1949: 246) Unlike Marovo, Hoava and Roviana clearly reflect the original use of subject markers, namely following markers of aspect/mood and preceding the verb, suggesting that such a structure was also present in their common ancestor language. It seems likely that in Proto Eastern New Georgia the negative particle also preceded the subject markers, and thus the verb complex structure in Table 9.3 can be reconstructed.7 The reconstruction of the verb complex structure in Table 9.3 considers the use of subject markers with the negative particle ka- in Marovo to reflect their original distribution. I propose that the use of subject markers with discourse connective particles also reflects this structure, but that in this case an original aspect/mood marker *ma ¼ has been reanalysed as the discourse connective particle ma ¼ ‘and then’. Table 9.4 sets out the stages of development which I propose have led to the use of preverbal subject markers with discourse connective particles in Marovo. Stage I shows that two original structures with a morpheme *ma are reconstructable; one in which *ma was a discourse connective particle and the other in which *ma ¼ was an aspect/ mood marker. Thus in one of the Stage I structures, *ma was a clause-initial discourse connective particle that preceded the verb complex. There is strong evidence that this morpheme and
Table 9.3. Posited structure of the verb complex in Proto Eastern New Georgia (ASPECT/ MOOD ¼ )
7
(NEGATIVE ¼ )
SUBJECT MARKER ¼
VERB ( ¼ OBJECT MARKER)
( ¼ DIRECTIONAL MARKER)
Further investigation is needed to confirm this analysis of negative clauses, particularly since subject markers can sometimes precede the negative particle in Roviana, and such an order is also found in western New Georgia languages.
174
Bethwyn Evans
Table 9.4. The development of the use of preverbal subject markers with discourse connective particles in Marovo Stage I
Constructions *ma
VERB COMPLEX
*ma ¼ SUBJ.M
II
*ma
VERB
VERB COMPLEX
*ma ¼ SUBJ.M
VERB
*ngina (ma ¼ SUBJ.M) III
*ma
VERB COMPLEX
*ma ¼ SUBJ.M
VERB
*ngina (ma ¼ SUBJ.M)
IV
*ma ¼ SUBJ.M
*ngina V
VERB
VERB
VERB COMPLEX
VERB
ma ¼ SUBJ.M
VERB COMPLEX
he ¼ SUBJ.M VERB COMPLEX boru( ¼ SUBJ.M) VERB COMPLEX pata( ¼ SUBJ.M) VERB COMPLEX
Function Discourse connective particle ma precedes clauseinitial verb complex Aspect/mood marker ma ¼ followed by subject marker and lexical verb Discourse connective particle ma precedes clauseinitial verb complex Aspect/mood marker *ma ¼ followed by subject marker and lexical verb Innovative epistemic adverb denoting possibility Discourse connective particle ma precedes clauseinitial verb complex Implied sequential meaning of aspect/mood *ma ¼ salient in certain contexts Gradual grammaticalisation of ngina such that it is extended to use in a broader range of contexts Two constructions with *ma merge; aspect/mood *ma ¼ is reanalysed as the discourse connective particle ma ¼ ngina used as a preverbal marker of irrealis mood Use of subject markers is extended to include cooccurrence with other discourse connective particles
From aspect/mood marker to discourse particle
175
construction, exemplified by (21) from Roviana, are reconstructable for Proto Oceanic (see Lynch et al., 2002: 89).8 Roviana (21)
[Nuquru la]VC [rau]ABS, me [nanas-i-u]VC [sa titisa]ERG and ask-TR-1SGO DEF:SG teacher enter go 1SG ‘I went in, and the teacher asked me . . . ’ (Corston, 1996: 31)
...
The other stage I structure comprises an aspect/mood marker *ma ¼ followed by a subject marker and the verb. This structure was present in Early Roviana, as shown by (18) and (22). Early Roviana (22) [Ma-qu tiok-i-a]VC [si asa]ABS. FUT-1SGS call-TR-3SGO abs 3SG ‘I will call her.’ (Waterhouse, 1926: 19) Lynch et al. (2002: 82–86) provide evidence supporting the reconstruction of the structure ASPECT/MOOD ¼ SUBJECT MARKER þVERB for Proto Oceanic, and suggest that there was likely an aspect/mood distinction between realis and irrealis, with irrealis as the morphologically marked category. Lynch (1975) presents comparative data supporting the reconstruction of *ma ‘irrealis mood’ for Proto Oceanic. However, Lynch et al. (2002: 85) note that the reconstruction of the forms of Proto Oceanic aspect/mood markers is exceedingly complex. The varying forms in contemporary languages seem to be the result of different, and sometimes repeated, processes of grammaticalisation which are difficult to reconstruct (see also Ross and Lithgow, 1989). At a lower level there is reasonably strong evidence to support the reconstruction of *ma ¼ for Proto Northwest Solomonic. While its exact function is difficult to determine, Ross (1982: 31) suggests a non-past/non-completive meaning. Proto Northwest Solomonic *ma ¼ is glossed here as ‘irrealis’ reflecting the use of its reflexes as markers of ‘irrealis’, ‘future’ and ‘imperative’. It should be noted that this reconstruction warrants further investigation. Stage II represents structures which are reconstructable at least back to the common ancestor of eastern New Georgia languages. An innovation at this stage is the epistemic adverb ngina, which denoted possibility. There is no direct evidence that ngina co-occurred with *ma ¼ , but the presence or absence of *ma ¼ in this construction does not affect the overall analysis. In Hoava,
8
Roviana has ergative-absolutive alignment of core noun phrases (see Corston, 1996), and in Roviana examples they are labelled ERG and ABS.
176
Bethwyn Evans
ngina is an epistemic adverb that occurs before the verb complex (Davis, 2003: 247–249), as in (23). Hoava (23) Ngina [koni tavet-i-a]VC [rao]SUBJ. FUT make-TR-3SGO 1SG possibly ‘I will possibly make it.’ (Davis, 2003: 249) Stages III and IV are those reconstructed for pre-Marovo. During this period I hypothesise that ngina underwent a gradual process of grammaticalisation; losing its specific epistemic meaning, extending in use to a broader range of contexts, and becoming a marker of irrealis mood. In contemporary Marovo ngina has a range of functions which can be described as irrealis, including future tense, (24) and habitual aspect, (25). Marovo (24)
‘‘[Ngina
tepaBtepa
IRR
RDPBpray
paki]VC first
[hita]SUBJ . . . ’’ 1PLINCL
‘‘We will pray first . . . ’’
(25)
[Manemaneke woman [ngina
ororeke wife taBtavete]VC IRR RDPBwork ‘My wife works in the garden all
ta-gu poss-1sgP pa LOC
raka]SUBJ, 1SG chigo garden
tongania every
kolokolo. time
the time.’
It is also argued that the two constructions with *ma, shown in stages I–III in Table 9.4, have merged in Marovo to a single construction with a clause-initial discourse connective particle ma ¼ that occurs with subject markers. Thus juxtaposed sequences of clauses with an initial mood marker ma-, like those in (26) from Early Roviana, have been reanalysed as clauses conjoined not by juxtaposition, but by the discourse connective particle ma ¼ , as are those in (27) from Marovo. Early Roviana boko taqarau]ABS (26) a. Uve, [ma-mu la va mate-a]VC [sa yes IMP-2SG go CAUS die-3SGO DEF:SG pig 1SGP b. [ma-da ganiBgani]VC. FUT-1PLINC RDPBeat a. ‘Yes, go and kill my pig’ b. ‘that we may eat.’ (Waterhouse, 1926: 18)
From aspect/mood marker to discourse particle
177
Marovo (27)
a. b. c. a. b. c.
[La]VC [raka]SUBJ pa LOC go 1sg ma ¼ [gu ngina choga]VC then ¼ 1SGS IRR jump ma ¼ [gu tope]VC. then ¼ 1SGS dive ‘I’ll go to the place for diving,’ ‘then I’ll jump in’ ‘and I’ll dive.’
tania here [raka]SUBJ 1SG
vasina place
tope-ani, dive-NOM
Stage V represents the contemporary Marovo system, where the two original structures with *ma have merged. The ma ¼ morpheme now has the discourse connective meaning, but the original morphology of the mood marker *ma ¼ , that is, the subject markers, has been retained and extended to all instances of the merged construction with ma ¼ . The use of subject markers has also been extended by analogy to constructions with other discourse connective particles.
5. ASPECT/MOOD MARKER
TO
DISCOURSE CONNECTIVE PARTICLE
The reconstruction of the reanalysis of an aspect/mood marker *ma ¼ as a discourse connective particle warrants further investigation in terms of motivations and processes of change. The change, proposed by the reconstruction in Table 9.4, is that the construction under (i) in Table 9.5 has been reanalysed as, and merged with, the construction under (ii). The linguistic evidence that such a change has occurred is the morphological parallels between the constructions in the form of subject marking. It is proposed that constructions like (i) in Roviana and like (ii) in Marovo represent the beginning and end of this change, which implies that certain processes of change have occurred in the history of Marovo. An important aspect of justifying the reconstruction presented in Table 9.4, and this change in particular, is demonstrating its nature and plausibility. The reanalysis proposed to have occurred here implies the diachronic reorganisation of both the syntactic and semantic composition of construction (i), and this paper highlights the need to consider the motivations and
Table 9.5. The two constructions with *ma (i) (ii)
ASPECT/MOOD MARKER DISCOURSE CONNECTIVE PARTICLE
¼ SUBJECT ¼ SUBJECT
MARKER MARKER
VERB VERB
Example (26) Example (27)
178
Bethwyn Evans
explanations of such changes from both these perspectives. The models of syntactic and semantic change, used here to inform the reconstruction, are described in Section 5.1. 5.1. Syntactic and semantic reanalysis Eckardt (2006: 106) comments that ‘‘[r]eanalysis rests crucially on specific communicative situations, the right kind of side message, the right kind of grammatical material’’. This view is reflected here by the fact that the syntactic motivations and explanations of reanalysis are viewed within the context of the communicative use of the construction, and semantic and pragmatic motivations and explanations are viewed within the context of morphosyntactic aspects of the construction. Harris and Campbell (1995) examine reanalysis from a syntactic perspective, mentioning only briefly the possibility of concurrent meaning change. Reanalysis, one of three mechanisms within their theory of syntactic change, is defined as change in ‘‘the underlying structure of a syntactic pattern . . . which does not involve any modification of its surface manifestation’’ (Harris and Campbell, 1995: 50, see also Langacker, 1977). Harris and Campbell (1995: 81–82) describe reanalysis and its actualisation in terms of three stages: Stage A, Input: The input structure has all of the superficial characteristics of the input analysis. Stage B, Actualization: The structure is subject to multiple analysis; it gradually acquires the characteristics of an innovative analysis, distinct from that of Stage A. Stage C, Completion: The innovative structure has all of the superficial characteristics of the innovative analysis. Reanalysis per se occurs at the transition between stages A and B that is the shift from one to more than one structural analysis of a construction. Indeed, Harris and Campbell (1995: 72) state that: the conditions necessary for reanalysis to take place are that a subset of the tokens of a particular constructional type must be open to the possibility of multiple structural analyses, where one potential analysis is the old one (applicable to all tokens) and the other potential analysis is the new one (applicable to a subset).
Stage B, the actualisation of reanalysis, often involves changes of extension, such that the construction undergoes changes in its usage and/or its surface manifestation that reflect the innovative structural analysis. From the perspective of reconstruction, it is the linguistic manifestation of Stage B changes that provides evidence of the underlying reanalysis. It should be noted that more than one structural analysis of a construction may be maintained over time, and some reanalyses may never reach Stage C.
From aspect/mood marker to discourse particle
179
Eckardt (2006: 236) characterises semantic reanalysis in a parallel way to Harris and Campbell (1995) for syntactic reanalysis: Semantic reanalysis: The process of semantic reorganization of a sentence whereby the salient overall conveyed information remains the same, but is composed in a different manner. What may have previously been in part assertion, in part implication, turns entirely into a literal meaning after reanalysis.
This view of semantic reanalysis places it within the context of models of meaning change that are pragmatically based. Within such models, semantic change is viewed to be dependent on and motivated by patterns of language use. For example, Traugott and Dasher (2002) set out the following diachronic path for meaning change: coded meaning W utterance-token meaning (invited inference) W utterance-type meaning (generalised invited inference) W new coded meaning (see also Levinson, 2000). Thus a shift in meaning begins where the use of a linguistic element (lexeme, construction) ‘invites’ an inferred meaning alongside its coded meaning in a particular context. Over time this inferred meaning may become a more salient aspect of the overall meaning of the linguistic element in this context, and be generalised to other contexts (generalised invited inference). A new coded meaning is seen to have developed once the linguistic element can be used in contexts where only the meaning of the generalised invited inference is appropriate. Hansen and Waltereit (2006) argue that pragmatically based meaning change is more complex than Traugott and Dasher’s (2002) single overarching pathway. Through a detailed examination of the different kinds of implicatures and their usage, they propose that it is more common for an utterance-token meaning (their particularised conversational implicature) to be directly semanticised as a new coded meaning. Eckardt (2006: 10) highlights the need to also consider the aspects of meaning change other than the semanticisation of a pragmatic inference, listing three factors that play a crucial role. First, for a construction to be reorganised, both semantically and syntactically, there needs to be the ‘right’ kind of match between the surface parts of the construction and the information conveyed. Second, Eckardt (2006: 10, Ch. 7) suggests that semantic reanalysis is restricted by semantic universals. Finally, the communicative characteristics of the onset contexts of a change must be different from the use of the construction in ordinary contexts, perhaps including that the inferred meaning occurs frequently relative to the use of the construction. 5.2. The reanalysis of *ma ¼ ‘irrealis mood’ in Marovo It is proposed here that the combination of syntactic and semantic/pragmatic factors motivated the reanalysis of construction (i) in Table 9.5 as construction (ii). From a syntactic perspective several factors appear to have motivated the development of a second structural analysis of
180
Bethwyn Evans
construction (i). First, structure (ii) was already a construction within the language (pre-Marovo), and the homophony of *ma ¼ ‘irrealis mood’ and *ma ¼ ‘and (then)’ meant there was surface similarity between the two underlying structures. Second, in Marovo, as well as in all New Georgia languages, clauses tend to be VSO, meaning that the sequence of a clause-initial discourse connective particle followed by the verb complex was not unusual. Third, comparative evidence suggests that not all clauses would have had overt aspect/mood marking, and that subject markers could occur as the sole preverbal element within the verb complex, thus increasing the occurrence of the sequence DISCOURSE CONNECTIVE PARTICLE þ SUBJECT MARKER þVERB. Finally, the development of an innovative irrealis marker, ngina, may have led to a decrease in the use of *ma ¼ as an aspect/mood marker in terms of at least some functions (e.g. future time reference). Thus the characteristics of pre-Marovo indicate that the ‘‘right kind of grammatical material’’ was present to allow for the two constructions in Table 9.5 to merge. The change to the underlying syntactic structure proposed to have occurred as part of this reanalysis is one of category labels (see Harris and Campbell, 1995: 61–65). That is, instances of *ma ¼ with the grammatical function of an aspect/mood particle came to have the possible analysis as a discourse connective particle. But what about ‘‘the right kind of side message’’? That is, was the overall conveyed information of construction (i) in Table 9.5 the same as that of construction (ii), at least for a subset of tokens? The use of ma- ‘imperative; future’ in Early Roviana, as described by Waterhouse (1949), appears to reflect at least some of the functions of Proto Northwest Solomonic *ma ¼ ‘irrealis’ and these are taken to be inherited functions in Roviana from the ancestor language of both Roviana and Marovo, Proto Eastern New Georgia. Thus the use of ma- in a corpus of Early Roviana narratives (Western Province Government, 1991) can be used to gain insights into the likely semantic and pragmatic interpretations of construction (i) in pre-Marovo.9 This corpus of Early Roviana comprises 19 narratives, all of which are traditional stories narrated in the 3rd person. In the 888-sentence corpus there are only 57 uses of the construction MA-SUBJECT MARKER þVERB, all of which are within direct speech. As expected the three functions of this construction described by Waterhouse (1949) are present in the corpus. The most frequent use of ma-, shown in (28b), was with the 2SG subject marker -mu and an imperative function. The other commonly occurring use of ma- was a hortative function, occurring with the 1PLINCL subject marker -da and the particle aria ‘call to action; come on’, as shown in (29). 9
These Early Roviana narratives, published by the Western Province Government of the Solomon Islands in 1991, were collected in the 1930s and 1940s. The narratives are published in Roviana only, and the grammatical and basic semantic analyses of examples presented here are based on Waterhouse’s (1949) Roviana dictionary and grammatical description, as well as grammatical description of contemporary Roviana (Corston, 1996; Corston-Oliver, 2002, 2003). Further analysis of semantic and pragmatic meanings of the examples is based on the context of the clause within the narrative as a whole.
From aspect/mood marker to discourse particle
181
Early Roviana (28) a. ‘‘Agoi tugo [kaqu ino va toa i-a]VC [sa nika]ABS, 2SG:FOC EMPH must make.fire CAUS live TR-3sgO DEF:SG fire b. [ma-mu va toa ngiBngira i-a]VC, . . . ’’ IMP-2SGS CAUS live RDPBstrong TR-3SGO a. ‘‘You must get the fire going, b. you make it strong, . . . ’’ (Western Province Government, 1991: 77) (29)
a. b. a. b.
‘‘Aria [ma-da peqo-i]VC [ka ngeta tomoko lavata]ABS’’, NUM three war.canoe large come.on IMP-1PLINCLS adze-TR [gua]VC [si asa]ABS. say ABS 3SG ‘‘Let’s carve (adze) three large war canoes’’, ‘he said.’ (Western Province Government, 1991: 26)
The imperative/hortative nature of ma- implies future time reference, which is otherwise not overtly indicated within the clause. From the context of (30b), for example, either a hortative or a future interpretation is possible. Within the corpus there are a few examples of ma- where the imperative/hortative meaning is not apparent, but the future time reference interpretation is, as in (31b). This future use of ma- typically occurs with a 1SG or 1PL subject participant, and indicates the speaker’s intention to carry out the event described. This again is a meaning which is also conveyed in the imperative clauses with ma-, that is the speaker intends that an event will be carried out and is instructing someone to do so. Early Roviana [Kera sing b. [ma-qu avavoso FUT-1SGS hear a. ‘‘ . . . Sing it again, b. ‘let me listen/I will
ko, pule n-ia]VC APPL-3SGO EXPL return qua]VC’’, gua se Eo. 1SGP say PERS:ABS brush.hen
(30) a. ‘‘ . . .
(31) a. Ba but b. ‘‘Arau 1SG:FOC c. pude if
[zama]VC speak [ma-qu FUT-1SGS [gore go.down
listen’’, said Brush Hen’ (Western Province Government, 1991: 50)
[se PERS:ABS
kopu-ni]VC watch-APPL mae]VC come
Viruviru]ABS, swordfish [sari DEF:PL [gamu]ABS 2PL
karua two kote soon
tomoko]ABS war.canoe, arau [pule la napo]VC’’, 1SG:FOC return go drink
182
Bethwyn Evans
d. [gua]VC [si asa]ABS say ABS 3sg a. But Swordfish said, b. ‘‘I will watch the two war canoes, c. when you (pl) go down, soon I will drink’’, d. he said (Western Province Government, 1991: 81) In summary, clauses with ma-SUBJECT MARKER in Early Roviana had coded meanings of imperative and hortative, and implied both future time reference and the speaker’s intention that an event be carried out. Examples like (31b) where the imperative/hortative meaning is not apparent, suggest that the meaning of future tense and speaker intention was also a coded meaning of clauses with ma-. In Marovo there are two uses of the construction ma ¼ SUBJECT MARKER. First, Marovo has a construction verb similar to Roviana in which ma ¼ occurs with the 1PLINCL subject marker da and the particle aria ‘come on’ and the construction has a hortative meaning, as in (32). In Marovo aria ‘come on’ can also be used alone with this same hortative meaning, as in (33). Marovo (32)
‘‘Aria ma ¼ [da come.on HORT ¼ 1PLINCLS ‘‘Let’s go to the bush, . . . ’’
kala go
la]VC go
pa LOC
goana, bush
. . . ’’
(33) [Jama]VC [Junior]SUBJ, ‘‘Aria, [raka]SUBJ mana [hiva omBomi tungana]VC’’ . . . talk J. come.on 1SG but want RDPBsee also ‘Junior said, ‘‘Let’s go, I want to have a look too’’ . . . ’ With the other use of the construction ma ¼ SUBJECT MARKER in Marovo ma ¼ is a discourse connective particle and indicates the temporal relationship between the events described by the preceding and following clauses. Although ma ¼ can be used to conjoin clauses describing simultaneous events, the most frequent function of ma- is to indicate a sequential relationship between the events; the event described by the following clause occurs after the event described by the preceding clause. For example, in (34), ma ¼ is used to indicate that the event ‘waiting for low tide’, (34a), was followed by the event ‘fishing with a net’, (34b). Marovo (34)
a.
Pata in.order
[vera wait
ni-na]VC TR-3SGO
[hoinWore oNOMWgo.down
ta POSS
mati]OBJ, shallow.reef
From aspect/mood marker to discourse particle
b.
ma ¼ [gu la talavuni vagara then-1SGS go start net ‘So I waited for low tide, and then I started to net.’
183
hua]VC. HUA
Despite these apparently very different meanings of the constructions with ma-SUBJECT MARKER in Early Roviana and Marovo, there are examples in each language where the ‘‘overall conveyed information’’ is very similar, if not the same. The Early Roviana example in (35) is from a traditional narrative about an earthworm and a centipede. This section of the narrative describes the earthworm telling the centipede to come to him early in the morning when he will tell about a medicine. While the first clause with ma-, (35a), has a clear imperative meaning, the second clause with ma-, (35b), describes the speaker’s intention to tell the addressee about the medicine, and conveys that the ‘telling’ event will occur after the ‘coming’ event described in the first clause. Thus, just as in Marovo, here in Early Roviana the construction with ma-SUBJECT MARKER conveys a sequential relationship between the events described by preceding and following clauses. The conveyed sequential relationship between events described by clauses with ma-SUBJECT MARKER is particularly apparent in Early Roviana, in cases, like (35), where there is a change in the subject participant and the event described in the second clause is conditional on the occurrence of the event described in the first clause. Early Roviana (35) a. [Zama]VC [se pilaka]ABS, ‘‘Vugo munumunu vaqavaqasa speak PERS:ABS worm tomorrow morning at.dawn sa rimata si [ma-mu mae]VC koa rau [lopu ele gasa]VC NEG yet leap DEF:SG sun FOC IMP-2SGS come to 1SG b. [ma-qu tozi ni-go]VC keke meresena hopena taqarau,’’ FUT-1SGS tell APPL-2SGO one medicine holy 1SGP [gua]VC [se pilaka]ABS. PERS:ABS worm say a. ‘Earthworm said, ‘‘Early tomorrow morning, before the sun has risen, you come to me.’ b. ‘I will tell you about my sacred medicine’’, said Earthworm’
(Western Province Government, 1991: 5) Even in contexts where the coded imperative meaning of a clause with ma-SUBJECT MARKER is clearly apparent, the conveyed sequential meaning may be an equally important part of the overall meaning of the clause. Example (36) is a series of imperative clauses where the speaker is instructing the addressee to climb a coconut palm and pick two drinking coconuts, dropping one down from the top of the palm and bringing the other down without dropping it. Thus it is not simply the imperative meaning which is important in the way in which this message is structured, but also that the overall event, climbing and getting two drinking coconuts, is done in a particular way, with each sub-event specified and to be carried out in the sequence that the speaker states.
184
Bethwyn Evans
Early Roviana (36) a. ‘‘Ke la [ma-mu haele i-a]VC thus go IMP-2SGS climb TR-3sgO [sa ngohara buma papaka-na hoi]ABS. DEF:SG coconut green short-3SGP that:SG b. [Ma-mu pakete vagi]VC [karua bulo]ABS. IMP-2SGS pluck get two drink.coconut c. Tamu goi si [ma-mu va hoqa i-a]VC pa pepeso. 2SG FOC IMP-2SGS CAUS fall TR-3SGO LOC land 2SGP d. Ba taqarau si [ma-mu paleke gore ni-a]VC but 1SGP FOC IMP-2SGS carry go.down APPL-3SGO [lopu va hoqa i-a]VC’’, [gua]VC. NEG CAUS fall TR-3SGO say a. ‘‘So go, you climb that short green coconut tree. b. You pick two drinking coconuts. c. Yours, you drop it to the ground. d. But mine, you carry it down, don’t drop it’’, he said. (Western Province Government, 1991: 62)10 While there are no examples within the Early Roviana corpus in which the imperative/hortative or future time reference interpretations are not available for a clause with ma-SUBJECT MARKER, there are only five examples with which a sequential meaning is clearly not appropriate. This can be contrasted with double that number of clauses in which the sequential meaning appears to be an equally important part of the meaning. In Marovo the construction with ma ¼ SUBJECT MARKER with the coded sequential meaning can, with some tokens, convey a hortative meaning. Thus in (37b) ma ¼ indicates that the event ‘starting work’ will follow that of ‘praying’. The context of this clause is also such that a hortative meaning is conveyed, though not indicated grammatically within the clause. Marovo (37)
a. b. a. b.
[ngina ‘‘[Mae]VC [hamu]SUBJ, come 2PL IRR beto ma ¼ [da ngina finish then ¼ 1PLINCLS IRR ‘‘You lot come, we’ll pray first, after that then we’ll start work’’.
tepaBtepa RDPBpray talavuni start
paki]VC first tavete]VC’’. work
[hita]SUBJ, 1PLINCL
10 In the published narratives paleke ‘to carry’ in (36d) is written as peleke. Since peleke is not in Waterhouse’s (1949) dictionary and the meaning of paleke ‘to carry’ is apparent from the context, I have taken this to be a typographical error.
From aspect/mood marker to discourse particle
185
From the Early Roviana examples in (35b) and (36) and the Marovo example in (37) it can be seen that there are tokens of the two different ma-SUBJECT MARKER constructions which have essentially the same ‘‘overall conveyed information’’, and it is these kinds of tokens of the constructions which must have provided the necessary conditions for the reanalysis. It seems likely that this change was triggered by the chance homophony of *ma ‘irrealis’ and *ma ‘and (then)’, and the surface similarity of the two constructions. The semantic/pragmatic aspects of the constructions would have also facilitated the change. It is proposed that the shift in meaning was from an utterance-token meaning to a coded meaning (cf. Hansen and Waltereit, 2006). This kind of detail is difficult to reconstruct, but examples in Early Roviana where the sequential meaning of constructions with ma- is most salient are within two specific contexts. First, where a future construction with ma- follows an imperative clause with ma-, as in (35) and the occurrence of the event of the second clause carried out by the speaker or speaker and addressee is dependent on the addressee carrying out the event denoted by the first clause. Second, in imperative clauses with ma- which specify a sequence of sub-events, as in (36). In other contexts the sequential meaning of clauses with ma- in Early Roviana is less salient and may reflect universal tendencies of narrative structure with respect to the expected temporal relationships. But did these two contexts occur frequently enough to facilitate a shift in coded meaning? Rather than frequency as such, it seems likely that it was a reduction in the use of *ma- ‘imperative/hortative; future’ which was crucial. In Marovo and in (Early) Roviana an imperative meaning can be indicated by bare verb stem, and it is possible that the ma- imperative in Early Roviana (and thus pre-Marovo) was used in contexts where the imperative nature of the clause was being emphasised, for example, when giving a list (or sequence) of instructions highlighting that the speaker expects the addressee to carry out each sub-event. In pre-Marovo ngina would have been another way of indicating future time reference, and so the use of ma ¼ with future time reference would be more likely to have occurred in contexts where an imperative/hortative interpretation was also possible. For example, contexts in which the addressee is required to carry out one event to allow the speaker or the speaker and the addressee to carry out a subsequent event. In this way, it seems possible that an overall decrease in the use of *ma ¼ ‘imperative/hortative; future’ may have resulted in a perceived frequency of the *ma ¼ SUBJECT MARKER construction in contexts where the sequential inference was particularly salient. The presence of *ma ‘and (then)’ would have facilitated the redistribution of meaning components within the construction, with *ma ¼ gaining a coded sequential meaning. Constructions (i) and (ii) with ma ¼ SUBJECT MARKER have merged in contemporary Marovo and now the discourse connective particle ma ¼ occurs obligatorily with the subject markers, even in cases where there is no coded or inferred meaning of imperative/hortative or of future time reference. The use of subject markers has also been extended to other discourse connective particles (see Section 3). The use of ma ¼ in Marovo in the construction aria ma ¼ da [come.on MA-1PLINCL] ‘let’s go’ is a relic of the original construction where *ma had a hortative function.
186
Bethwyn Evans
5. CONCLUDING REMARKS In summary, it was the combination of morphosyntactic as well as semantic and pragmatic characteristics of both construction (i), ASPECT/MOOD ¼ SUBJECT MARKER þVERB, and construction (ii), DISCOURSE CONNECTIVE PARTICLE VERB COMPLEX, which motivated the reanalysis described here. The presence of both constructions in pre-Marovo which in some contexts would have conveyed the same overall information (coded and inferred meanings), provided the onset conditions for the change. While much of the reconstruction of processes of change remains speculative, it can be seen that the combined use of established models of syntactic and semantic change as well as the detailed comparison of the usage of constructions within the contemporary languages allows for a greater understanding of the development, and thus explanation, of synchronic structures.
ACKNOWLEDGMENTS The research for this paper has been supported by a Simon Fellowship at the University of Manchester. This fellowship and also a British Academy Small Grant (SG-40401) provided financial support for the fieldwork carried out on Marovo, both of which are gratefully acknowledged. An earlier version of this paper was presented at the 18th International Conference on Historical Linguistics in Montre´al (2007), and I would like to thank the audience there for their comments. I would also like to thank Luisa Miceli and Louise Mycock for comments on an earlier version of this paper.
REFERENCES Bourchier, L. (2007). Vangunu Report. Unpublished manuscript. Canberra: Australian National University. Corston, S. H. (1996). Ergativity in Roviana, Solomon Islands. Canberra: Pacific Linguistics. Corston-Oliver, S. (2002). ‘‘Roviana’’, in J. Lynch, M. Ross and T. Crowley (eds.), The Oceanic languages. Surrey: Curzon, Richmond, 467–497. Corston-Oliver, S. (2003). ‘‘Core arguments and the inversion of the nominal hierarchy in Roviana’’, in J. W. Du Bois, L. E. Kumpf and W. J. Ashby (eds.), Preferred argument structure. Grammar as architecture for function. Amsterdam: John Benjamins, 273–300. Davis, K. (2003). A grammar of the Hoava language, Solomon Islands. Canberra: Pacific Linguistics. Eckardt, R. (2006). Meaning change in grammaticalization. An enquiry into semantic reanalysis. Oxford: Oxford University Press. Erteschik-Shir, N. (2007). Information structure. The syntax-discourse interface. Oxford: Oxford University Press.
From aspect/mood marker to discourse particle
187
Evans, B. (2008). ‘‘Subject agreement in Marovo: Diachronic explanations of synchronic conditions’’. Oceanic Linguistics 47 (2): 383–408. Hansen, M.-B. M. and R. Waltereit (2006). ‘‘GCI theory and language change’’. Acta lingvistica hafniensia 38: 235–268. Harris, A. C. and L. Campbell (1995). Historical syntax in cross-linguistic perspective. Cambridge: Cambridge University Press. Langacker, R. W. (1977). ‘‘Syntactic reanalysis’’, in C. N. Li (ed.), Mechanisms of syntactic change. Austin, TX: University of Texas Press, 59–139. Levinson, S. (2000). Presumptive meanings. The theory of generalized conversational implicature. Cambridge, MA: MIT Press. Lynch, J. (1975). ‘‘Oral/nasal alternation and the realis/irrealis distinction in Oceanic languages’’. Oceanic Linguistics 14: 87–99. Lynch, J., M. Ross and T. Crowley (eds.) (2002). The Oceanic languages. Surrey: Curzon Press, Richmond. Ross, M. (1982). ‘‘The development of the verb phrase in the Oceanic languages of the Bougainville region’’, in A. Halim, L. Carrington and S. A. Wurm (eds.), Papers from the Third International Conference on Austronesian Linguistics Vol. 1: Currents in Oceanic. Canberra: Pacific Linguistics, 1–57. Ross, M. (1988). Proto Oceanic and the Austronesian Languages of western Melanesia. Canberra: Pacific Linguistics. Ross, M. (in press). ‘‘Lexical history in the Northwest Solomonic languages’’, in J. Bowden and N. Himmelmann (eds.), A journey through Austronesian and Papuan linguistic and cultural space. Canberra: Pacific Linguistics. Ross, M. D. and D. R. Lithgow (1989). The prehistory of some western Oceanic tense/mood markers: Insights from natural morphosyntax. Unpublished manuscript. Canberra: Australian National University. Strawson, P. F. (1964). ‘‘Identifying reference and truth values’’. Theoria 30: 86–99. Traugott, E. C. and R. B. Dasher (2002). Regularity in semantic change. Cambridge: Cambridge University Press. Waterhouse, J. H. L. (1926). A Roviana phrase book. Sydney: Epworth Press. Waterhouse, J. H. L. (1949). A Roviana and English dictionary. With English-Roviana index, list of natural history objects and appendix of old customs. Sydney: Epworth Printing and Publishing House. Western Province Government (1991). Na buka vivinei malivi pa zinama Roviana (Roviana Custom Stories Book). Solomon Islands: Western Province Government.
Current Trends in Diachronic Semantics and Pragmatics Edited by Maj-Britt Mosegaard Hansen and Jacqueline Visconti r 2009 by Emerald Group Publishing Limited. All rights reserved.
10 THE GRAMMATICALIZATION CHANNELS OF EVIDENTIALS AND MODAL PARTICLES IN GERMAN: INTEGRATION IN TEXTUAL STRUCTURES AS A COMMON FEATURE Gabriele Diewald, Marijana Kresic and Elena Smirnova
1. INTRODUCTION This paper is concerned with German evidentials and German modal particles (henceforth: MPs). Both serve as linguistic means for expressing different grammatical contents in present-day German (henceforth: PDG), and both have developed via the process of grammaticalization. Sentences (1)–(4) illustrate the PDG usage of the infinitive constructions with werden ‘become’, scheinen ‘seem’, drohen ‘threaten’ and versprechen ‘promise’ as grammatical markers of evidentiality; sentences (5)–(8) illustrate the prototypical PDG usage of the MPs aber, ja, eben, denn, serving here as examples for the class of MPs as a whole. Evidential constructions: (1) (2) (3) (4)
Peter wird jetzt im Bu¨ro sein. ‘Peter will be in the office right now.’ Sie scheint ihren Schirm im Bu¨ro gelassen zu haben. ‘She seems to have left her umbrella in the office.’ In Ungarn drohen die Da¨mme zu brechen. ‘In Hungary the dams threaten to break.’ Der Abend verspricht ein Erfolg zu werden. ‘The evening promises to be a success.’
190
Gabriele Diewald et al.
Constructions with MPs: (5) (6) (7) (8)
Das ist aber keine gute Idee. ‘That is ABER not a good idea.’ Es soll ja auch schwimmen. ‘It is JA meant to float.’ Deutsch ist eben schwer. ‘German is EBEN difficult.’ Warum fahren Sie denn so schnell? ‘Why are you DENN driving that fast?’
The two categories in question differ to a great extent in their function as well as in their formal realization. Evidentiality is concerned with indicating the information source the speaker is relying on to make a claim. This places this category next to epistemic modality without, however, merging them into one.1 Evidential distinctions are realized in German by periphrastic verbal constructions, i.e. what we witness is the grammaticalization of auxiliaries and verbal periphrastic forms which is a well-known general tendency in the German language (cf. Section 2.1). As to the class of MPs, their function is a discourse grammatical one: an MP marks the utterance containing it as a non-initial turn. This is achieved by relating the proposition to a pragmatically presupposed unit, i.e. to the discourse context, to the hearer’s/the speaker’s shared knowledge and assumed thoughts.2 MPs constitute a formal word class of their own, with peculiar clustering characteristics, the development of which is no less than the rise of a new, previously unknown grammatical category in German (cf. Section 2.2). Thus, the two categories at first sight do not seem to have more in common than being categories that have not received much attention by linguists as far as the German language is concerned. In this paper, we will show that common developmental tendencies may be postulated in the grammaticalization of even such functionally different linguistic elements as evidential markers and discourse grammatical devices like MPs. The intention behind this approach is to propose that these common features are potentially powerful indicators of grammaticalization channels in general.
1
Evidentiality has been often subsumed under the area of epistemic modality (e.g. Palmer, 1986; Ifantidou, 2001). In our view, however, evidentiality – as a semantic domain as well as a grammatical category – has to be kept notionally distinct from other categories, in particular from modality and the grammatical category of mood (cf. Anderson, 1986; de Haan, 1999, 2001b; Aikhenvald, 2003, 2004). 2 It should be noted that the speaker by using a modal particle – and contrary to a particle-less but otherwise identical utterance – claims the existence of such a context. This claim may or may not be in line with the actual situation.
The grammaticalization channels of evidentials and modal particles in German
191
Before discussing the details of the two grammaticalization channels, some introductory remarks are in place. Linguistic change accruing into grammaticalization is in principle functional change which usually affects every linguistic level (i.e. phonological, morphological, syntactic, semantic etc.). For a full account of a specific grammaticalization channel, it is necessary to attend to each layer and describe its behaviour in the process. In this paper, however, we will not be able to do this for reasons of space. Instead, we will concentrate mainly on the basic functional changes that entities undergo during grammaticalization. For this purpose, we look at two developmental tendencies, which are known to be relevant for grammaticalization in general, and – as we contend – which are analogous for the grammaticalization of evidentials and MPs in German: While the phenomenon of scheme retention (a) is a kind of prerequisite and therefore will not receive particular attention here, the idea of the successive semantic-functional stages in grammaticalization (b) is central for our topic. 1.1. a. The tendency of scheme retention and reinterpretation elaborated by Sweetser (1990) Sweetser (1990) claims that in the semantic change accompanying grammaticalization processes, we can make out a basic semantic template (Sweetser calls it ‘image scheme’) that is retained in its basic relational structure through all developmental stages. In the course of semantic change, this relational template is successively transferred to different semantic domains. These semantic domains are characterized, on the one hand, by increasing abstraction. On the other hand, they also provide new semantic features that had not been present before. Sweeter’s model has been successfully employed in the investigation of various grammaticalization processes in German (cf. Diewald, 1999b for modal verbs; Diewald, 2006 for MPs; Smirnova, 2006 for the verb wu¨rde). As will be shown below, both categories, evidentials as well as MPs, develop along the reinterpretation and abstraction of a relational semantic template. In advanced stages of grammaticalization, this development results in the indexical–grammatical interpretation of that template. 1.2. b. The model of successive semantic–functional stages suggested by Traugott (1989) and developed further by Traugott and Dasher (2002) Traugott’s well-known findings on general tendencies of semantic change (accompanying grammaticalization) have been demonstrated to be relevant for many categories in a number of languages. We argue that beyond the change in the semantic domains involved, which is the major focus of Traugott’s model, these tendencies indicate a functional change concerning the semiotic status of the item in question (including word class variation). In our view, it is this change of semiotic status which is the essence of grammaticalization.
192
Gabriele Diewald et al.
The development runs from mainly referential to mainly textual and connective functions, and further evolves to indexical grammatical functions.3 Therefore, we use Traugott’s model with some minor shifts of focus and adaptations (e.g. collapsing the external and internal situation into one function, i.e. into the ‘referential function’). The development, as we understand it, may be sketched as follows:4 (i) (self-contained) referential function W(ii) text-integrative/connective function W(iii) indexical-grammatical function In the following, the intermediate stage from lexical to grammatical will receive particular attention, which Traugott calls ‘‘Tendency II’’ associating it with the rise of a ‘‘textual’’ function of the element (cf. Traugott 1989, 1995, 1997). In her 1989 paper, she describes this tendency as follows: ‘‘meanings based in the external or internal described situation Wmeanings based in the textual and metalinguistic situation’’ (Traugott, 1989: 34f ). We will show that this stage, i.e. the shift of the relational template into the textual plane and the reinterpretation of a purely referential relational meaning into a meaning connecting linguistic elements and thus providing textual cohesion, is an essential step in the development of evidentials as well as MPs in German. Furthermore, it will be shown that there are astonishing similarities in the specific characteristics of this stage in both categories. By way of generalization, we may assume that the stage at which a linguistic element comes to be very closely integrated into textual structures can be found (to more or less extent) in every grammaticalization process. The subsequent sections of the paper are organized in the following manner. In the second section, we discuss the functional status of both – evidential constructions and MPs – in PDG, arguing for treating them as grammatical markers. In the third section, we present common
3
For a discussion of the notions of indexicality and subjectivity see Diewald (1999b: 14–16). Traugott’s full scheme was given in Traugott (1989: 34f). In subsequent studies (e.g. Traugott, 1999; Traugott and Dasher, 2002) this model was expanded, refined and generalized in order to account for as many instances of semantic change as possible. The modifications concern, first, the notion of subjectification, which is complemented by ‘‘intersubjectification’’ in order to account for social deixis (Traugott and Dasher, 2002: 23f), and second, the integration of possibly concomitant structural changes, especially the change of scope and topological features in the development of discourse markers (Traugott, 1999: 177ff). The chronological ordering of the three stages as well as the description of the respective semantic and cognitive domains has been retained in its essence. 4
The grammaticalization channels of evidentials and modal particles in German
193
grammaticalization channels along which German evidential constructions and MPs have developed. The last section summarizes our findings.
2. THE FUNCTIONAL STATUS OF EVIDENTIALS IN PRESENT-DAY GERMAN
AND
MODAL PARTICLES
Claiming that German evidential expressions and MPs are grammatical markers, i.e. that they represent grammatical categories in PDG, this section briefly summarizes how we define this function and the respective oppositions, without, however, going into extensive argumentation on this issue (for this, see Diewald, 1997, 2006, where the grammatical status of the MPs is argued for, and Diewald and Smirnova, to appear, for the evidential constructions). There are two criteria we consider here to be pivotal for defining a grammatical sign. These are (i)
the indexical potential of a sign, and
(ii)
its paradigmatic integration.
Indexical potential refers to the specific relational content of a sign (cf. Diewald, 2006: 414ff).5 A grammatical sign modifies another (lexical) sign by relating it to some other element, i.e. to some reference point lying outside both of them. That is, a grammatical sign establishes a link between the linguistic element it modifies and some other entity. Apart from this indexical potential, a grammatical sign usually displays some (abstract and schematic) semantic content that specifies/ denotes the established link. The relational component of grammatical signs can be visually represented as a vector and illustrated in the following diagram (taken from Diewald, 2006: 415): point of reference’(grammatical sign & unit modified by grammatical sign) The paradigmatic integration of a sign, or its paradigmacity refers to the fact that the members of a grammatical paradigm, which as a whole is constituted by a categorical value, are closely linked to each other, whereby we often find intra-categorial sub-groupings (cf. Lehmann, 1982/1995). It can be noted here that the two criteria described above may be attributed to one essential property of a grammatical sign, i.e. to its inherently relational character. In the first case, the syntagmatic perspective on the relational character of a sign is taken: a grammatical sign always combines with (an)other lexical sign(s), and this complex syntagmatic unit is related to a particular point of reference. In the second case, the paradigmatic view on the relational character of a sign is
5
The fact that ‘indexicality’ is a defining feature of grammatical elements was first mentioned by Jespersen (1921). In a similar vein, Jakobson (1957/1971) considers a substantial group of grammatical signs to belong to a particular type of signs, i.e. ‘indexical symbols’, whereby ‘‘shifters’’ constitute a sub-class of them.
194
Gabriele Diewald et al.
predominant: every grammatical sign is very closely related to other grammatical signs, so that they always participate in relatively closed and highly structured grammatical paradigms. Needless to say, there are more criteria than these two that could be adduced as grammaticality properties of a sign. However, as we believe, the two features mentioned above are the most prominent ones. Therefore, we will focus on them in the following analysis of the functional status of German MPs and evidential constructions. 2.1. German evidential periphrases as grammatical elements The four constructions scheinen & zu & infinitive, drohen & zu & infinitive, versprechen & zu & infinitive and werden & infinitive, serve as inferential evidential periphrases in PDG (cf. Diewald and Smirnova, to appear, for further discussion). In this function, they mark the proposition as inferred by the speaker from some other facts. These can be either observable facts, or some otherwise perceived information, or reported evidence, or even general knowledge of the speaker etc. The indexical potential of evidential constructions can be shown by contrasting two clauses which only differ from each other in the existence of an evidential auxiliary in one of them: (9)
Die Lieferung ist gro¨Xer als erwartet. ‘The delivery is larger than expected.’ (10) Die Lieferung scheint/droht/verspricht gro¨Xer zu sein als erwartet. ‘The delivery seems/threatens/promises to be larger than expected.’
Sentence (9), which has no evidential marker, is a factual claim that is neutral to its communicative context: it does not make any reference to some other linguistic or non-linguistic entity (leaving aside the referential function of the article and the indexical elements of the finite verb). Sentence (10), on the other hand, contains additional information that the speaker has some evidence for making this factual claim. Scheinen & zu & infinitive marks that the proposition is inferred from some pieces of information which the speaker experiences (e.g. s/he sees the delivery) or has experienced previously (e.g. s/he has read the documentation sent beforehand). Drohen/versprechen & zu & infinitive indicate that the proposition is inferred from some currently accessible pieces of information (e.g. s/he is observing the delivery or the documentation right now). The evidential auxiliaries are indexical insofar as they relate the proposition to the information available to the speaker. Hence, an evidential construction links the proposition in which it occurs to a non-linguistic point of reference, which is the information source of the speaker. The indexical character of the grammatical category of evidentiality has been recognized first by Roman Jakobson (1957/1971) and further explored in recent studies on evidentiality (see, e.g. Floyd, 1996; Diewald, 2001; de Haan, 2001a; Faller, 2002; Smirnova, 2006). Evidentials relate the proposition to the information source, this information source being inherently linked to the speaker, i.e. it is incorporated within the speaker’s (deictic) position. So, it is the information
The grammaticalization channels of evidentials and modal particles in German
195
presented as available to (and by) the speaker which is the defining property of evidentiality. The abstract conceptual structure of evidential markers may be noted in the following manner: (speaker’s) information source’(evidential construction & proposition) As shown in this scheme, the (speaker’s) information source serves as the point of reference to which the proposition is linked. The specific value of inferential evidentials consists in specifying the relation between the proposition and the information source as inference, i.e. the proposition is the result of the speaker’s inference based on some evidence. Hence, the German infinitive constructions, all carrying inferential evidential meaning, meet the first criterion of grammatical signs we have proposed in the pervious section: they have indexical potential. As far as the second criterion is concerned, the four constructions pertain to one (sub-)domain of evidentiality – to the domain of inferential evidentiality. This is the common semantic basis they all share with each other. Formally, they are all verbal periphrastic constructions with infinitives. Within the subcategory of evidentiality, each of the four constructions specifies its common semantic basis in a different way. That is, each construction has a distinctive feature within the paradigm, which are structured as (privative) oppositions. This means that the four constructions form four levels of conceptual organization; each lower level is defined by the presence of an additional distinctive feature. The neighbouring levels form binary oppositions, differing from each other only with regard to the presence or absence of one distinctive feature. Werden & infinitive is the unmarked member of the paradigm. Diachronically, it is the eldest member of the paradigm. It usually indicates that the speaker’s inference is based on some evidence or general knowledge, cf.: (11)
Peter wird jetzt im Bu¨ro sein. ‘Peter will be in the office right now.’ [e.g. the speaker sees light in Peter’s office windows, or knows that Peter is always in the office at this time, or s/he just called him at home and Peter didn’t answer the phone, etc.]
Scheinen & zu & infinitive is marked as opposed to werden & infinitive as it normally means that the inference process is based on some specific pieces of information which has been in some way acquired (i.e. observed, heard from others etc.) by the speaker: (12)
Sie scheint ihren Schirm im Bu¨ro gelassen zu haben. ‘She seems to have left her umbrella in the office.’ [e.g. the speaker sees her entering the room dripping wet; or s/he has been called by a friend telling that she had returned to the office and would be late for dinner, etc.]
Drohen and versprechen & zu & infinitive are in opposition to the construction with scheinen (as well as with werden) in that they specify the basis of inference as present experience. This
196
Gabriele Diewald et al.
means that the speaker bases his/her inference on some specific, presently accessible (i.e. observed) pieces of information. Diachronically, these two constructions are the youngest evidential constructions within the German inferential evidential paradigm. (13) (14)
Wenn ein Haus aus den Na¨hten zu platzen droht. (DWDS, Kernkorpus) ‘When a house threatens to burst at the seams.’ . . . und nach la¨ngerem Suchen und nachdem er ihn mit den Za¨hnen getestet hatte, fand Ku¨renberg einen Reis, der ko¨rnig zu kochen versprach. (IDS-Corpora, W) ‘ . . . and after a lengthy search and after testing it with his teeth, Ku¨renberg found a type of rice which promised to boil grainy.’ [e.g. the speaker sees the first observable traces of bursting at the house; or s/he is tasting something that suggests that the rice will boil grainy]
Hence, the grammatical paradigm made up of the four German constructions is constituted by one categorical value, which is inferential evidentiality. Moreover, the members of the paradigm are closely linked to each other by oppositions.6 In sum, the four infinitive constructions constitute a (close) grammatical paradigm, i.e. the German evidential periphrases meet also the second criterion defining a grammatical sign. 2.2. Modal particles as grammatical elements As explicated above, the defining features for the status as a grammatical sign used here are (i) the indexical potential of a sign and (ii) its paradigmatic integration. Both features can be found in the class of MPs today. The indexical meaning of MPs has long been recognized by many scholars.7 Its specific nature can be shown by contrasting two utterances which only differ from each other in the existence of a
6
It should be noted here that, while describing how the German evidential paradigm is organized, we have considered only purely evidential values associated with the four constructions in question. As is known, however, the constructions with drohen and versprechen display additional semantic components. First, they are used to introduce events which are not realized yet, i.e. they are future oriented. Second, while drohen normally introduces negatively evaluated (i.e. non-desired) situations, versprechen is used to introduce positively evaluated (i.e. desired) situations. These components may be traced back to the semantics of the lexical source verbs and are still preserved in their grammaticalized variants. As drohen and versprechen are the least grammaticalized members of the German evidential paradigm (see Diewald and Smirnova, to appear), it is not surprising that they still show preservations of the original semantics, which – as we assume – will slowly fade away in the further grammaticalization. 7 Hentschel (1986: 31), for example, regards the ‘metacommunicative deixis’ of the MPs as their constitutive meaning. Franck (1979: 8) holds that the MPs have a sort of anaphoric function, and goes on to specify the element the MP is pointing to as the ‘‘specific premises [ . . . ] about the argumentative and interactional context of the utterance’’ [our translation]. A similar view is expressed by Aijmer et al. (2006) and Fischer (2006).
The grammaticalization channels of evidentials and modal particles in German
197
MP in one of them, cf.: (15) (16)
Deutsch ist ‘German is Deutsch ist ‘German is
schwer. difficult.’ eben schwer. EBEN difficult.’
Sentence (15), which has no MP, represents an unmodalized statement that is maximally independent of or neutral to its communicative context. It does not make any reference to some other linguistic or non-linguistic entity (ignoring the referential function of the NPs and the deictic elements of the inflected verb, which of course must be present in any finite sentence). The MP eben in sentence (16), on the other hand, adds the information that the speaker regards the proposition Deutsch ist schwer as given, as communicatively understood. The MP is indexical insofar as it points backwards from the linguistic unit in which it appears and relates the utterance to a proposition or speech-act alternative the speaker regards as relevant and given, i.e. the MP relates the utterance to a kind of pragmatic presupposition. This indexical relational structure has been shown to be common to all MPs (Diewald, 1997, 2006). It is given in the following scheme: pragmatically given unit’(MP & utterance in the scope of the MP) As to the question why this relational structure should be regarded as grammatical meaning the following points should be considered. In referring ‘back’ to something that is treated as communicatively given, albeit unexpressed, the MP marks the utterance which contains it as noninitial. As the interchange of initial and responsive turns is the constitutive feature of spoken interaction, this relational function of the MPs clearly is an indispensable grammatical device for structuring discourse. In other languages, the embedding of an utterance in the preceding discourse is achieved with the help of other linguistic means, such as idiomatic phrases and other lexical chaining, intonation and discourse markers (e.g. tag questions in English). In German, the speaker makes use of MPs in order to mark a turn as non-initial and responsive, and thus is able to manipulate and modify the ongoing interchange. Two further points must be noted here: first, the proposition to which the modalized utterance refers is usually not explicitly expressed in the preceding linguistic material. Second, the content of the proposition which the speaker treats as given and refers back to is expressed in the modalized utterance for the first time. Its propositional core is essentially identical with the core of the modalized utterance; through the MP, it is modified in some communicatively relevant way.8
8
This view is similar to Foolen’s description of the class-constitutive meaning of modal particles: ‘‘Als Klassenbedeutung fu¨r Modalpartikeln gilt, daX sie immer auf eine implizite, im Kontext relevante Proposition hinweisen. Diese implizite Proposition ist immer eine logische Variante der explizit ausgedru¨ckten Proposition’’ (Foolen, 1989: 312f).
198
Gabriele Diewald et al.
As to the paradigmatic organization of MPs it may suffice to note the following. In reference grammars as well as in linguistic literature, MPs are defined as a word class that is constituted by the clustering of specific formal, structural and functional features.9 Adopting this position and taking into consideration the indexical relation described above, we treat MPs as a grammatical paradigm characterized by the following features:
non-inflecting linguistic elements/words,
syntactically integrated,
appear only in the middle field of the sentence,
have no constituent value,
relate propositions and speech-act alternatives, one of which is not textually expressed but treated as ‘given’.
While these features make for the external separation from other word classes, the internal paradigmatic organization of the class of MPs is comparable to the internal organization of other grammatical paradigms. In addition to their class constitutive indexical meaning, each MP has a diachronically motivated, lexeme-specific semantic feature. In the case of the MP eben, this distinctive semantic feature consists in specifying the relation as an iterative one, i.e. it states a repetition of the pragmatically given proposition in the present scene. On the other hand, the specific semantic content of aber is ‘adversative’, and that of ja is ‘affirmative’, to give just two more examples here. The following sentences show this; added to them is a fragmentary explication of the particular distinctive semantics called ‘adversative’, ‘affirmative’ and ‘iterative’ here (for details see Diewald and Fischer, 1998; Diewald, 2007): (17)
(18)
(19)
Das ist aber keine gute Idee. ‘That is ABER not a good idea.’ [adversative with respect to the proposition: ‘That is a good idea – someone may think this; it is not true’] Das ist ja keine gute Idee. ‘That is JA not a good idea.’ [affirmative with respect to the proposition: ‘That is not a good idea – you and me have known this before’] Das ist eben keine gute Idee. ‘That is EBEN not a good idea.’ [iterative with respect to the proposition: ‘That is not a good idea – I held this opinion before and I hold it and say it now’]
As MPs are sensitive towards speech-acts and sentences types, ‘minimal pairs’, such as the ones given above, in most cases apply only to subgroups of the whole class. The examples 9
The core of this class consists of the following 15 extremely frequent items: aber, auch, bloX, denn, doch, eben, eigentlich, etwa, halt, ja, mal, nur, schon, vielleicht, wohl (Gelhaus, 1995: 371; Helbig and Buscha, 2001: 421ff).
The grammaticalization channels of evidentials and modal particles in German
199
demonstrate minimal pairs in statements, i.e. representative speech acts. As mentioned in Section 2, the organization into subparadigms is a common feature of grammatical paradigms, so that this fact does not constitute a counterargument against treating MPs as a grammatical paradigm. To conclude: both, evidentials and MPs function as indexical signs, i.e. they link linguistic elements to some relevant point of reference. Both, evidentials and MPs, take scope over propositions. While evidentials relate the proposition to the speaker’s information source, MPs relate it to a pragmatically given unit. However, there exist differences in the paradigmatic organization of the denotative meaning of evidentials and MPs which will not be further discussed here.
3. THE GRAMMATICALIZATION CHANNELS PARTICLES (IN GERMAN)
OF
EVIDENTIALS
AND
MODAL
3.1. Evidential constructions In Diewald and Smirnova (to appear) we have proposed a common developmental path for the four German evidential constructions. Table 10.1 sketches the three successive diachronic stages in the grammaticalization from their lexical source semantics via the intermediate stage of a construction with textually relational meaning towards the expression of inferential evidentiality (for a detailed, corpus-based diachronic analysis see Diewald and Smirnova, to appear). As shown in Table 10.1, the first type of textual structures, at which some of the four German inferential evidential constructions participated during their grammaticalization, manifest the successive or consecutive relation between two (explicitly expressed) events. As the following example shows, the source aspectual semantics of the verb werden in the Old and Middle High German periods indicates a strongly related continuum between the beginning of the event and the event itself. (20)
Thoˆ sliumo uuard thar mit themo engile menigi himilisches heres got lobontiu inti quedentiu: . . . (KALI-Corpus, Tatian 6,3; 9th century) ‘suddenly the crowd was there with the angel praising god and saying: . . . ’
At the second stage (which approximately covers the Early New High German period), i.e. in the constructions with werden & infinitive or present participle, its lexical semantics is reinterpreted as an indication of the successive relation between two (different) situations which are textually connected to each other. The following example illustrates this: (21)
Und wie der knecht ungetrew Auch sterben muost, das ward ir new. Und ward darumb wainen ser, . . . (Kauringer Gedichte, XIV, 635, 14th century, taken from Aron, 1914) ‘And as the disloyal servant also had to die, that was new to her. And therefore she started to cry bitterly . . . ’
200
Gabriele Diewald et al.
Table 10.1. Successive stages in the development of the German evidential constructions 1st stage: referential function
2nd stage: textintegrative/ connective function
Ingressive aspectual verb
Speech act verb
Verb of visual effect
werden
drohen/versprechen
scheinen
Successive/consecutive relation between two textually expressed events
Simultaneous/ comparative relation between two events
werden & infinitive & pres. part.
scheinen & zu & infinitive & adjective & adverb
drohen/versprechen & zu & infinitive & accusative object
3rd stage: indexicalgrammatical function
Inferential evidential
werden & infinitive
drohen/versprechen & zu & infinitive
scheinen & zu & infinitive
At this stage, werden is integrated in larger textual configurations profiling a successive relationship between two situations without actually being a means of constructing or specifying such structures, i.e. werden is not a grammatical element, but the grammatical reading is one of several alternative interpretations of the whole textual structure. Finally, at the third stage, including only the construction werden & infinitive, inferential evidential meaning develops, marking primarily a logical connection between two situations, whereby one of them is regarded as evidence for the other one, but is no longer explicitly represented in the text: (22)
Peter wird jetzt im Bu¨ro sein. ‘Peter will be in the office right now.’
For drohen and versprechen, a similar developmental path can be assumed. First, being speech act verbs, until the beginning of the 19th century, they served to announce an event which was intended
The grammaticalization channels of evidentials and modal particles in German
201
by the subject and thus introduce a connection between two events, i.e. between the announcement itself and the announced action. This is the case in the examples (23) and (24): (23)
(24)
. . . gab die feurige Eleonora endlich selbst Gelegenheit, daX ich meine hefftigen Seuffzer und Klagen kniend vor derselben ausstieX, und mich selbst zu erstechen drohete, woferne . . . (Bibliotheca Augustana, Schna4, 18th century) ‘ . . . finally, passionate Eleonora herself gave me a chance to kneel down in front of her and heave my heavy sighs and complaints and to threaten to stab myself, if . . . ’ . . . und dess habend sich die von zu¨rich gelopt und versprochen vir sy und jr nachkomen war und stet zu halten wie optstat. (BonnerFru¨hneuhochdeutschKorpus, 213, 15th century) ‘ . . . and the people from Zurich vowed and promised for themselves and their offspring to keep this true and for ever . . . ’
At the second stage (ca. 19th–20th century), in constructions with accusative objects or with infinitives, these verbs come to mark a successive relationship between two situations. Similarly to werden, they are integrated in larger textual configurations without being grammatical markers. Here, their speech act semantics is already largely bleached and only the relational connection between two events is promoted, whereby one event textually precedes the other. (25)
(26)
Spuren der Plu¨nderung lieXen sich bemerken in Gefolg innerer Feindschaft. hohe Mauern drohten den Einsturz, Tu¨rme standen unsicher. (IDS-Corpora, GOE/ AGA.01784, 19th century) ‘ . . . high walls threatened the collapse/to collapse . . . ’ Offenbach am Main zeigte schon damals bedeutende Anfa¨nge einer Stadt, die sich in der Folge zu bilden versprach. (IDS-Corpora, GOE/AGD.06345, 19th century) ‘Even then, Offenbach am Main displayed significant features of that town, into which it promised to develop in future.’
Finally, the textually represented connection between the events turns into the inferential evidential meaning, presenting a logical link between situations one of which is seen as evidence for another. Importantly, the pieces of evidence are not mentioned in the text. Instead, they are implied as available to the speaker by the use of the evidential construction with drohen or versprechen. This still ongoing process can be witnessed in PDG: (27) (28)
Wenn ein Haus aus den Na¨hten zu platzen droht. (DWDS, Kernkorpus) ‘When a house threatens to burst at the seams.’ . . . und nach la¨ngerem Suchen und nachdem er ihn mit den Za¨hnen getestet hatte, fand Ku¨renberg einen Reis, der ko¨rnig zu kochen versprach. (IDS-Corpora, W) ‘ . . . and after a lengthy search and after testing it with his teeth, Ku¨renberg found a type of rice which promised to boil grainy.’
202
Gabriele Diewald et al.
Scheinen, a perceptual verb coding visual effect, implies that some entity triggers a particular effect by means of its appearance. Until the late 16th century, this is its main function. (29)
Bluot skein in uuangoˆn . . . (KALI-Corpus, Ludwigslied, 9th century) ‘Blood shone in cheeks . . . ’
From the late 16th century onward, this verb in combination with adjectives, adverbs and infinitives comes to indicate a relation between two situations which are compared to each other on the basis of their appearance. In such cases, scheinen very often occurs in textual configurations in which two different entities or events are compared. Together with (traditional) comparative means such as als ‘as, than’ and wie ‘as’, scheinen participates (though not truly actively) at presenting such structures. (30)
es sind noch einige kleinigkeiten u¨brig, welche so nu¨zlich seyn als sie gering scheinen. (Bibliotheca Augustana; Leib 1, 17th century) ‘there are still some small things left, which are as useful as they seem trivial.’
At the third stage, beginning in the 18th century, in construction with infinitives, scheinen develops towards an inferential evidential, indicating that the described event is inferred from another one which is in some way similar to it and thus serves as piece of evidence: (31)
Sie scheint ihren Schirm im Bu¨ro gelassen zu haben. ‘She seems to have left her umbrella in the office.’
By way of conclusion, we can state that, first of all, the four German evidential periphrases arise via grammaticalization processes developing from main verbs towards auxiliary verbs. Their overall semantic development is an exemplary case of the successive reinterpretation of an abstract relational template, i.e. the interplay of persistence (namely that of the abstract structure) and change (namely its concrete ‘‘fillings’’) through time (see Section 1). Furthermore, in their grammaticalization, they exhibit an intermediate stage at which they are integrated in specific textual structures without actively participating at constructing them. As grammaticalization proceeds, they gradually become grammatical markers that (actively) modify propositions due to their grammatical – inferential evidential – function. At the same time, their integration in particular text structures becomes looser and looser till they are not longer identified with specific contextual configurations.
3.2. Modal particles For various reasons, it is not yet possible to present more than a first rough sketch of a possible grammaticalization channel for the category of MPs. First, the class of MPs comprises a much
The grammaticalization channels of evidentials and modal particles in German
203
larger number of elements than, e.g. the class of evidentials. Second, the ultimate source lexemes and source constructions of MPs display a considerable span of variation – ranging from ‘original’ particles with other than modal function (e.g. ja), adverbs (schon, OHG scono, adv. of adjective sconi), adjectives (eben, gleich, ruhig), to two-word syntagms (ni ware Wnur, vil lihte Wvielleicht) etc. (cf. Diewald, 1997, 2006). Third, there do not exist diachronic studies for every particle so far. Table 10.2 presents a sketch of a possible grammaticalization channel of MPs. We have structured it analogous to the scheme suggested for the evidentials (see Section 3.1) in order to demonstrate the similarities of the grammaticalization of both categories. The three particles denn, aber and eben used here serve as examples, representing the variation of the class as a whole.10 The development of these particles displays all the factors mentioned in Section 1 as relevant for grammaticalization channels. On the one hand, the items develop in accordance with Sweetser’s (1990) model encompassing the reinterpretation and abstraction of a relational semantic template and resulting in the indexical interpretation of that template. On the other hand, the items have a relational semantic structure that is reinterpreted according to the stages set up by Traugott.11 Furthermore, the development of the textual stages resembles closely the one found in the evidentials. While denn and aber at stage II represent successive relations among textual segments, eben represents simultaneous textual relations (comparison between two items present in the textual universe at the same time). Deriving from an old Indo-European demonstrative particle (1st stage), dann/denn has had indexical value all the time and was used in a MP function as early as in the MHG period (since 13th century). Still, in order to be used as a MP, it had to pass the stage as a temporal and text connective (consecutive) adverbial (2nd stage), which – placed in the middle field and restricted to questions – received the reinterpretation as a MP (3rd stage).12 As a MP, denn indicates a consecutive relation between the pragmatically given unit and the relevant situation. It marks the speech act as a non-initial, reactive turn. Aber, which derives from a comparative form of a local, i.e. directional adverb, meaning ‘farther away’ (1st stage), developed an adversative and copulative/iterative meaning, whereby the relational structure of the comparative ‘farther away’ was reinterpreted as ‘and then’, or ‘and something other’ etc.
10
For details of their development see Hentschel (1986), Diewald (1999a), and Autenrieth (2002). There are many specific details and peculiarities in the grammaticalization of modal particles, e.g. the question of word class constitution, word order etc. These will not be treated here, as we are concerned with the commonalities between evidentials and modal particles. For a thorough discussion see Diewald (1997, 2006). 12 For details concerning the diachronic development of denn/dann see Wegener (2002). 11
204
Gabriele Diewald et al.
Table 10.2. Successive stages in the development of the German modal particles 1st stage: referential function
2nd stage: textintegrative/ connective function
3rd stage: indexicalgrammatical function
Old demonstrative particles/adverbs
Adverbs with relational semantics
Adjectives with relational semantics
denn Origin in local demonstrative
aber Origin in comparative of local adverb
eben Origin in relational spatial, concrete adjective
Successive/consecutive Relation between two textually expressed events/instances
Simultaneous/ comparative relation between two events/ instances
denn & interrogative speech act
eben & so & two comparanda
aber & proposition
Indicating non-initial state of utterance, i.e. discourse grammatical value & particle-specific meaning
denn as MP in interrogative speech acts: consecutive meaning
aber as MP in exclamative speech acts: adversative meaning
eben as MPs in statements: iterative meaning
At the second stage, it developed conjunctional and adverbial usages and displayed a wide range of text connective functions, like the following one: (32)
der eine [ging] in sıˆn dorf, abir der andere zuo sıˆme gewerbe. (Evangelienb. Behaim 51 B. (1343), DWB Neubearb. 1, 181) ‘One of them [went] to his village, ABIR the other to his trade.’
As example (32) shows, even the conjunctional use need not be strictly adversative, but can express mere juxtaposition or even a copulative relation between the two propositions. As a
The grammaticalization channels of evidentials and modal particles in German
205
conjunctional element, abir clearly has text-connective functions, expressing an adversative/ copulative relation between two propositions that are textually present. The rise of this function corresponds to ‘Tendency II’ in Traugott’s model. Finally, from this developed the anchoring in the pragmatic presupposition (Diewald, 1999a). Aber reaches the third stage of Traugott’s model as soon as it is used as a MP. This new function is first documented in the 18th century.13 It is illustrated by the following example: (33)
Ihr mu¨Xt aber hier ja¨mmerliche Langeweile haben. (GoeWb, taken from Paul/Henne, 1992) ‘You must ABER be terribly bored here.’
Here, aber shows neither referential nor text-connective/conjunctional functions. Instead, it relates the utterance to its pragmatic pretext (see Section 3), marking it as non-initial and expressing a slightly adversative meaning. It has reached the stage of the PDG MPs. As the particles originate in very different structures, the details display a range of variation, e.g. the 2nd stage of eben, which derives from an adjective, differs distinctly from that of aber, as eben never was used as a conjunction. Instead, its text connective function was instantiated by comparative structures, where it was used to reinforce comparative conjunctions. Eben, which acquired its MP function in the Early New High German period (around 1500), in PDG, like all other MPs, is a linguistic device with the central function to refer back to something communicatively given, marking the utterance containing it as noninitial. In short, all MPs investigated so far in their diachronic development show the interplay of general grammaticalization scales concerning the semantic development, the functional development and the inherent semantic core structure of the items in question. The core structure remains stable in its essence, but is reinterpreted for its broad semantic or functional domains, which become more and more abstract, less referential and more grammatical. This semantic change, or more precisely this shift of core structure to various domains, is accompanied by a functional change corresponding to the general tendencies set up by Traugott, which transcends word-class boundaries. The result is the coexistence of several heterosemes whose functional spectra, though seemingly unrelated in a purely synchronic perspective, retain the successive gradient steps of regular grammaticalization.
13
The details of this development cannot be discussed here; see Diewald (1999a, 1997) for an extensive treatment.
206
Gabriele Diewald et al.
4. CONCLUSION By applying Traugott’s (Traugott, 1989; Traugott and Dasher, 2002) model of successive stages in semantic change (accompanying grammaticalization) to the developments of evidentials and MPs in German, this paper has worked out the following suggestions:
German evidential expressions and MPs are grammatical markers, i.e. they represent grammatical categories in PDG. Both display two essential characteristics of a grammatical sign: (i) they are indexical and (ii) they show a paradigmatic integration.
Evidentials and MPs are two functionally and formally distinct grammatical categories in German. The category of inferential evidentiality in German is realized by periphrastic verbal constructions, which indicate the information source the speaker is relying on to make a claim. MPs, on the other hand, constitute a formal word class of their own, displaying a discourse grammatical function: a MP marks the utterance containing it as a non-initial turn.
Three successive diachronic stages can be detected in the grammaticalization of the German evidential verbal periphrases: (i) from their lexical source semantics via (ii) the intermediate stage of a construction with textually relational meaning towards (iii) the expression of inferential evidentiality. As we have shown, they exhibit an intermediate stage at which they are integrated in specific textual structures (without actively participating in constructing them).
The grammaticalization of MPs can be also divided into three stages: from (i) particle or member of other word class with demonstrative or relational meaning, via (ii) a means for marking a relation between two textually expressed events/instances, towards (iii) grammatical markers indicating the non-initial state of an utterance. The three particles denn, aber and eben served as example cases to show that the development of the textual stages resembles closely the one found in the evidentials. While denn and aber at stage (ii) represent successive relations among textual segments, eben represents simultaneous textual relations.
Traugott’s – slightly adapted – model of semantic change applies to the grammaticalization of German evidentials as well as to the grammaticalization of MPs, since the members of these two grammatical categories go through the following developmental stages: (i) referential function W(ii) text-integrative/connective function W(iii) indexical-grammatical function. As we have demonstrated, the development of a text-integrative or text-connective function is essential for both, the grammaticalization of German evidentials and MPs. By way of conclusion, we propose that this intermediate stage, which is characterized by the establishment of a text-connective function, is highly relevant for grammaticalization processes in general. In this respect, our developmental scheme, focusing mainly on functional-semiotic
The grammaticalization channels of evidentials and modal particles in German
207
changes a sign undergoes in grammaticalization, is compatible with the Traugott’s model of semantic change. Finally, it should be noted that our findings require further empirical validations which can only be achieved by conducting extensive corpus-based investigations of different grammaticalization phenomena.
REFERENCES Aijmer, K., A. Foolen and A.-M. Simon-Vandenbergen (2006). ‘‘Discourse particles in the perspective of heteroglossia’’, in K. Fischer (ed.), Approaches to discourse particles. Amsterdam: Elsevier, 100–114. Aikhenvald, A. Y. (2003). ‘‘Evidentiality in typological perspective’’, in A. Y. Aikhenvald and R. M. W. Dixon (eds.), Studies in evidentiality. Amsterdam: Benjamins, 1–31. Aikhenvald, A. Y. (2004). Evidentiality. Oxford: Oxford University Press. Anderson, L. B. (1986). ‘‘Evidentials, paths of change, and mental maps: Typologically regular asymmetries’’, in W. L. Chafe and J. Nichols (eds.), Evidentiality. Norwood, NJ: Ablex, 273–312. Aron, A. W. (1914). Die ‘‘progressiven’’ Formen im Mittelhochdeutschen und Fru¨hneuhochdeutschen. Frankfurt am Main: Baer & Co. Autenrieth, T. (2002). Heterosemie und Grammatikalisierung bei Modalpartikeln. Eine synchrone und diachrone Studie anhand von eben, halt, e(cher)t, einfach, schlicht und glatt. Tu¨bingen: Niemeyer. de Haan, F. (1999). ‘‘Evidentiality and epistemic modality’’. Southwest Journal of Linguistics 18: 83–101. de Haan, F. (2001a). ‘‘The place of inference within the evidential system’’. International Journal of American Linguistics 2: 193–219. de Haan, F. (2001b). ‘‘The relation between modality and evidentiality’’, in Linguistische Berichte, Special Vol. 9, Synchronie und Diachronie der Modalverben. Hamburg: Buske, 201–216. Diewald, G. (1997). Grammatikalisierung. Eine Einfu¨hrung in Sein und Werden grammatischer Formen. Tu¨bingen: Niemeyer. Diewald, G. (1999a). ‘‘Die Entwicklung der Modalpartikel aber: ein typischer Grammatikalisierungsweg der Modalpartikeln’’, in H. O. Spillmann and I. Warnke (eds.), Internationale Tendenzen der Syntaktik, Semantik und Pragmatik. Akten des 32. Linguistischen Kolloquiums in Kassel 1997. Frankfurt am Main: Lang, 83–91. Diewald, G. (1999b). Die Modalverben im Deutschen: Grammatikalisierung und Polyfunktionalita¨t. Tu¨bingen: Niemeyer. Diewald, G. (2001). ‘‘Scheinen-Probleme: Analogie, Konstruktionsmischung und die Sogwirkung aktiver Grammatikalisierungskana¨le’’, in Linguistische Berichte, Special Vol. 9, Synchronie und Diachronie der Modalverben. Humburg: Buske, 87–110. Diewald, G. (2006). ‘‘Discourse particles and modal particles as grammatical elements’’, in K. Fischer (ed.), Approaches to discourse particles. Amsterdam: Elsevier, 403–425. Diewald, G. (2007). ‘‘Abto¨nungspartikel’’, in L. Hoffmann (ed.), Handbuch der deutschen Wortarten. Berlin: De Gruyter. Diewald, G. and K. Fischer (1998). ‘‘Zur diskursiven und modalen Funktion der Partikeln aber, auch, doch und ja in Instruktionsdialogen’’. Linguistica 38: 75–99.
208
Gabriele Diewald et al.
Diewald, G. and E. Smirnova (to appear). ‘‘The German evidential constructions and their origins: A corpus based analysis’’, in W. de Mulder, J. Mortelmans and T. Mortelmans (eds.), Papers presented at the International Conference on Tense, Aspect, Mood, and Modality, 18–20 September 2006, University of Antwerp, Belgium. Faller, M. (2002). Semantics and Pragmatics of Evidentials in Cuzco Quechua. PhD thesis, Stanford University, Department of Linguistics. Fischer, K. (2006). ‘‘Frames, constructions, and invariant meanings: The functional polysemy of discourse particles’’, in K. Fischer (ed.), Approaches to discourse particles. Amsterdam: Elsevier, 427–448. Floyd, R. (1996). ‘‘The radial structure of the Wanka reportative’’, in E. H. Casad (ed.), Cognitive linguistics in the Redwoods: The exploration of a new paradigm in linguistics. Berlin: Mouton de Gruyter, 895–941. Foolen, A. (1989). ‘‘Beschreibungsebenen fu¨r Modalpartikelbedeutungen’’, in H. Weydt (ed.), Sprechen mit Partikeln. Berlin: De Gruyter, 305–317. Franck, D. (1979). ‘‘Abto¨nungspartikeln und Interaktionsmanagement. Tendenzio¨se Fragen’’, in H. Weydt (ed.), Die Partikeln der deutschen Sprache. Berlin: De Gruyter, 3–13. Gelhaus, H. (1995). ‘‘Die Wortarten’’, in Duden: Grammatik der deutschen Gegenwartssprache. 5th Edition. Mannheim: Dudenverlag, 85–398. Helbig, G. and J. Buscha (2001). Deutsche Grammatik. Ein Handbuch fu¨r den Ausla¨nderunterricht. Berlin: Langenscheidt. Hentschel, E. (1986). Funktion und Geschichte deutscher Partikeln: ‘‘Ja’’, ‘‘doch’’, ‘‘halt’’ und ‘‘eben’’. Tu¨bingen: Niemeyer. Ifantidou, E. (2001). Evidentials and relevance. Amsterdam: Benjamins. Jakobson, R. (1957/1971). Shifters, verbal categories, and the Russian verbThe Hague: Mouton. 130–147. Jespersen, O. (1921). Language: Its nature, development, and origin. London: MacMillan. Lehmann, C. (1982/1995). Thoughts on grammaticalization. Mu¨nchen: Lincom Europa. Palmer, F. R. (1986). Mood and modality. Cambridge: Cambridge University Press. Paul/Henne 1992 ¼ Paul, H. (1992). Deutsches Wo¨rterbuch. 9., vollsta¨ndig neu bearbeitete Auflage von Helmut Henne und Georg Objartel unter Mitarbeit von Heidrun Ka¨mper-Jensen. Tu¨bingen: Niemeyer. Smirnova, E. (2006). Die Entwicklung der Konstruktion wu¨rdeþInfinitiv im Deutschen. Berlin: De Gruyter. Sweetser, E. (1990). From etymology to pragmatics: Metaphorical and cultural aspects of semantic structure. Cambridge: Cambridge University Press. Traugott, E. C. (1989). ‘‘On the rise of epistemic meanings in English: An example of subjectification in semantic change’’. Language 65: 31–55. Traugott, E. C. (1995). ‘‘Subjectification in grammaticalization’’, in S. Wright and D. Stein (eds.), Subjectivity and subjectification. Cambridge: Cambridge University Press, 31–54. Traugott, E. C. (1997). ‘‘Subjectification and the development of epistemic meaning: The case of promise and threaten’’, in T. Swan and O. J. Westvik (eds.), Modality in Germanic languages. Berlin: Mouton de Gruyter, 185–210. Traugott, E. C. (1999). ‘‘The rhetoric of counter-expectation in semantic change: A study in subjectification’’, in A. Blank and P. Koch (eds.), Historical semantics and cognition. Berlin: Mouton de Gruyter, 177–196. Traugott, E. C. and R. Dasher (2002). Regularity in semantic change. Cambridge: Cambridge University Press. Wegener, H. (2002). ‘‘The evolution of the German modal particle denn’’, in G. Diewald and I. Wischer (eds.), New reflections on grammaticalization. Amsterdam: Benjamins, 379–394.
The grammaticalization channels of evidentials and modal particles in German
CORPORA KALI-Corpus:http://www.kali.uni-hannover.deaccessed on 2107.2007 BonnerFru¨hneuhochdeutschKorpus:http://www.ikp.uni-bonn.de/dt/forsch/fnhd/accessed on 0511.2006 Bibliotheca Augustana:http://www.fh-augsburg.de/Bharsch/augustana.htmlaccessed on 0303.2007 IDS-corpora:http://www.ids-mannheim.deaccessed on 0706.2006 DWDS-Corpushttp://www.dwds.deaccessed on 08,062006
209
Current Trends in Diachronic Semantics and Pragmatics Edited by Maj-Britt Mosegaard Hansen and Jacqueline Visconti r 2009 by Emerald Group Publishing Limited. All rights reserved.
11 EVIDENTIALITY, EPISTEMICITY, AND THEIR DIACHRONIC CONNECTIONS TO NON-FACTUALITY Mario Squartini
1. INTRODUCTION The synchronic relationship between evidentiality and epistemic modality is far from being a settled issue. On the one hand, extensive typological surveys (Aikhenvald, 2004) have confirmed that in some of the world’s languages, possibly concentrated in given linguistic areas (de Haan, 2005), evidentiality is expressed by a separate set of dedicated morphemes that should not be confused with the less systematic coexistence of evidential and epistemic meanings described, among others, in Romance and Germanic modals (Dendale, 1994; Mortelmans, 2000; de Haan, 2001). On the other hand, the same typological direction of research has shown that even bona fide evidentials may have ‘‘epistemic extensions, to do with probability and speaker’s evaluation of the trustworthiness of information’’, as in Estonian, where the reported evidential ‘‘has an overtone of doubt’’, while ‘‘[v]isual evidential in Quechua can refer to information the speaker vouches for’’ (Aikhenvald, 2007: 212; see also Aikhenvald, 2004: 179–185). Commenting on these examples, Aikhenvald (2007: 226, footnote 3) correctly points out that ‘‘[t]he presence of such extensions does not make evidentials into ‘modals’ (contrary to some assumptions)’’, but it cannot be denied that these extensions contribute to blurring the distinction between evidentiality and epistemicity. Eventually, this may provide evidence in favor of those general models in which the interaction of the two categories plays a major role (Plungian, 2001). In this respect, diachronic data become crucial as a test bed of semantic connections among different domains of
212
Mario Squartini
the epistemic–evidential interface, a point that this article intends to pursue providing examples from well-attested Romance diachronic data.
2. DIACHRONIC CORRELATIONS
BETWEEN
EVIDENTIALITY AND EPISTEMICITY
The very fact that evidentials have epistemic, possibly pragmatic, extensions suggests a diachronic reappraisal that should be most crucially concerned with the interactions between evidentiality and epistemicity and should ultimately test the directions of diachronic developments. Anderson’s (1986) general reconstructive picture, where different diachronic data are put together, shows that the semantic evolution of evidentials may be accompanied by a gradual downgrading of the epistemic commitment. In Anderson’s semantic map (1986: 284) this is suggested by the diachronic evolution of ‘weak’ inferential forms (‘seem’) that may develop out of more strongly assertive circumstantial inferentials (‘appear’). Thus, the epistemic overtones described by Aikhenvald might not only be synchronic pragmatic extensions, some of them being also stable diachronic semantic developments. More generally, the idea that evidential meanings are primitives from which epistemic extensions are diachronically derived is also what the theoretical generalizations in Traugott (1989) and Traugott and Dasher (2002), based on subjectification as the primary force in semantic and pragmatic change, would predict. Being rooted on external direct (visual, auditory, etc.) or indirect sources, evidentials are intrinsically more objective than epistemics, the latter crucially involving the subjective commitment of the speaker to the factuality of the situation. Now, an obvious question arises: do we have cases demonstrating the opposite diachronic tendency (from epistemicity to evidentiality)? Aikhenvald mentions the possibility, apparently less widespread, that ‘modals’ develop into evidentials (2004: 276–279, 284–285), which suggests that the Romance reportive Conditional may also be representative of the tendency of modal markers to become evidentiality strategies (Aikhenvald, 2007: 213). But, in fact, the empirical coverage on such diachronic evolutions is still too sparse to be a reliable basis for more general assumptions. We probably know enough about the original sources of evidentiality (sensory evidentiality typically derives from the reanalysis of verbs of perception, while verbs of speech may be reanalyzed as quotatives, see Anderson, 1986 and Aikhenvald, 2004: 271–275), but we still ‘‘need to know what intermediate stages of grammaticalization of evidentials look like’’ (Aikhenvald, 2004: 287). In order to fill this gap extensive diachronic research would be required. Far from providing conclusive evidence in this respect, what follows is more modestly intended to show that the diachronic picture may be more complicated than so far assumed. It will be claimed that even a double-layered semantic model, including epistemicity and evidentiality, is not sufficient to cover the diachronic evolution of forms having evidentiality as a (relevant) part of their meanings. In particular, it will be shown that a significant, and often neglected, element is represented by the notion of non-actuality or
Evidentiality, epistemicity, and their diachronic connections to non-factuality
213
non-factuality,1 which has to be intended as the fictitious creation of non-factual worlds, not to be confused with epistemic subjective judgments (inference, conjectures). Different Romance diachronic evolutions, based on pragmatic and semantic extensions of inflectional forms and lexical items, will be compared to show that non-factuality is crucially intermingled with the semantic evolution of evidentiality.
3. STARTING
WITH A
DIACHRONIC DILEMMA
The list of Romance verb forms that can be described as evidentials is still disputed, but all those who strove to find evidential features in the Romance verb system seem to agree in recognizing an evidential (reportive) meaning in the Conditional. This was the first Romance inflectional form extensively studied by Dendale (1993) from the point of view of the interaction between evidentiality and epistemicity and is still recognized as basically evidential in Pietrandrea (2005), just to mention a first seminal work and a more recent and comprehensive reappraisal. In these analyses the Conditional is considered ‘evidential’ whenever it signals that the information is based on an indirect, external source (second-hand news or hearsay), as is shown in the following Italian example: (1)
Secondo le ultime notizie il presidente si sarebbe dimesso According to the latest news the president resigned (lit. ‘would have resigned’)
In Aikhenvald’s (2004) view this can be considered as an ‘evidentiality strategy’, a pragmatic extension that coexists with the basic non-evidential meaning of the form. As mentioned above, such evidential extensions seem to parallel the epistemic overtones that Aikhenvald recognizes in several evidential systems (see now Aikhenvald, 2007: 231), thus suggesting that these processes can be bidirectional. In any case, assuming that such semantic extensions can be considered as diachronic developments and even limiting the analysis to this single case, it seems reasonable to elaborate on the diachronic relationship between reportive evidentiality and other, possibly more
1
Along with non-factuality, it might also be worth considering the role of deontic modality as far as its diachronic connections to evidentiality are concerned. As shown in Ziegeler (2003) and Visconti (2004: 181–186), who describe the evolution of English be supposed to, evidential meaning may also foster a deontic interpretation as an intermediate stage towards more subjective epistemicity. The diachronic link between deontic modality and evidentiality can also be observed in Traugott’s (1989: 41–42) data on the semantic evolution of the English modal should, whose Old English forerunner was also a reportive evidential.
214
Mario Squartini
basic, meanings of the Romance Conditional. (2) represents the usage of the Italian Conditional as a marker of non-factual modality: (2)
Se potessi, andrei al cinema If I could, I would go to the movies
The hypothesis that the non-factual context in (2) can be considered as the trigger of the evidential usage in (1) has been put forward on several occasions and most consistently by Korzen and Nølke (2001: 133) as far as the French Conditional is concerned. Nonetheless, additional data show that this relationship might be too simplistic, if one considers that the Conditional also occurs as future in the past in Italian (3), as in the other Romance languages: (3)
Annuncio` che si sarebbe dimesso il giorno successivo He announced that he would resign (lit. ‘would have resigned’) on the following day
As (3) shows, the future in the past occurs, among others, in complement clauses depending from verba dicendi, a context that is in itself reportive and can easily trigger the independent evidential usage as reportive marker represented in (1). As for the French Conditional, the synchronic relationship between the occurrence in reported speech and the evidential (reportive) usage has already been noticed by Caudal and Vetters (2005: 120). Moreover, a diachronic process of desubordination of speech complements giving rise to reportive evidentials has been described among others in Estonian (see Aikhenvald, 2004: 281–283), which suggests a possible diachronic source for the reportive Conditional. As a result of the present discussion, the reportive usage emerges with two possible sources, in principle both equally reasonable, the non-factual meaning in (2) and the future in the past in (3). There are, however, synchronic data that may shed light on this diachronic dilemma, if the French Conditional is also taken into account. The interesting contrastive point between French and Italian (Squartini, 1999) is that from a semantic point of view the Conditionals of the two languages are totally overlapping, while their morphosyntax is significantly different. In Italian only the compound form of the Conditional can occur in future in the past contexts (3), where the simple Conditional is ungrammatical (4). On the contrary, a simple Conditional is required in French (5) in such contexts: (4) (5)
*Annuncio` che si dimetterebbe il giorno successivo He announced that he would resign on the following day Il annonc- a qu’il donnerait sa de´mission le lendemain He announced that he would resign on the following day
In French the compound Conditional may also be grammatical, provided the right aspectual context, as is the case in (6), where the situation is anterior with respect to a reference time (avant midi). This triggers a ‘perfect’ reading of the compound form that aspectually contrasts with (5),
Evidentiality, epistemicity, and their diachronic connections to non-factuality
215
where the temporal adverb (le lendemain) denotes the time when the event is expected to take place and not the reference time: (6)
Je croyais qu’il aurait fini son travail avant midi I thought he would have finished his work before noon
It can be concluded that in French it is anteriority that requires the compound Conditional, while in Italian the Compound Conditional is mandatory in whatever aspectual configuration of the future in the past. This behavior of the French Conditional can be capitalized on with respect to the puzzling interpretation of the reportive usage mentioned above. When used with reportive meaning the Conditional does not prove to be aspectually sensitive, the French Compound Conditional being generalized in any past contexts, irrespective of aspectual distinctions. This is demonstrated by (7), where the adverb de´ja` forces the interpretation of the reportive Conditional as anterior, while in (8) the adverb hier ‘yesterday’ temporally locates the event and in both cases the same form, a Compound Conditional, occurs: (7)
(8)
Aux dernie`res informations, le pre´sident aurait de´ja` donne´ sa de´mission ‘According to latest news the president has already resigned (lit. ‘would have already resigned’)’ Aux dernie`res informations, le pre´sident aurait donne´ sa de´mission hier ‘According to latest news the president resigned (lit. ‘would have resigned’) yesterday’
Conversely, the simple form of the reportive Conditional is required when referred to situations located in the present or future (9): (9)
Aux dernie`res informations, le pre´sident donnerait sa de´mission maintenant/demain ‘According to the latest news the president will resign (lit. ‘would resign’) now/tomorrow’
Unlike the future in the past discussed above (5–6), in which the distinction simple-compound was demonstrated as aspectually sensitive, the reportive Conditional (7–9) proves itself as sensitive to temporal distinctions. Such a morphosyntactic divergence of the reportive Conditional as opposed to the future in the past suggests that the latter is not likely to be the direct source of the former. On the contrary, the morphosyntax of the non-factual Conditional is more consistent with the behaviour of the reportive Conditional. Not differently from the reportive Conditional, whose simple form is restricted to present and future situations (9), the non-factual Conditional also excludes the simple form with past temporal reference, only admitting it in the present or future: (10)
Si je pouvais, j’irais au cine´ma (maintenant, demain, *hier) ‘If I could, I would go to the movies (now, tomorrow, *yesterday)’
216
Mario Squartini
Due to their morphosyntactic similarities, the most probable solution of our initial question is that the evidential meaning derives from the non-factual Conditional. In what follows other evidence will be presented supporting the hypothesis of a diachronic relationship between nonfactuality and evidentiality. Intriguingly, these additional data, taken from the originally pragmatic and eventually semantic evolution of a lexical item, will show that not only can non-factuality be the trigger of the evidential meaning but also the eventual step in the development of an originally evidential marker. This confirms the role of non-factuality, also suggesting that the diachronic tendency from non-factuality to evidentiality can be reversed (from evidentiality to nonfactuality).
4. FROM EVIDENTIALITY SPANISH DIZQUE
TO
NON-FACTUALITY: THE CASE
OF
AMERICAN
The semantic evolution from evidentiality to non-factuality can be clearly detected in the American Spanish particle dizque, diachronically derived from the gradual decategorialization of a form of the verb decir ‘say’ followed by the complementizer que. As the following Mexican data from Olbertz (2007: 155, 163) demonstrate, the complementizer may lose its subordinating function leading to the reanalysis of the construction as a modifier at different clause and phrase levels,2 including nominal modification (12): (11)
(12)
[at work, a nanny talks about her own children] Siempre tuvieron celos, dizque ma´s me ocupaba de e´stos que de ellos ‘They have always been jealous, saying I cared more for these children than for them’ les juro, compan˜eros, que todos aquellos dizque catedra´ticos que se hayan visto involucrados en el bochornoso caso de Rogelio [ . . . ] sera´n destituidos de sus cargos. ‘I swear to you, colleagues, that all those so-called professors who have been involved in the embarrassing Rogelio case [ . . . ] will be removed from their posts.’
As noted by Aikhenvald (2007: 220) adverbs or particles such as these ‘‘can be considered evidentials in the making – akin to evidential strategies’’, which goes beyond the scope of lexical evidentiality for they encroach into the grammatical domain. Leaving aside more general interpretative issues on the categorization of these forms, including their double function both as adverbs and particles discussed in Olbertz (2007), what is mostly relevant for the purpose of the present article is the semantic interpretation of dizque. Travis (2006) and Olbertz (2007)
2
See Olbertz (2007) for a revealing discussion on the interplay between syntactic functions and semantic interpretations.
Evidentiality, epistemicity, and their diachronic connections to non-factuality
217
independently and consistently demonstrate that, despite the transparent origin from a verb of saying, dizque is not restricted to evidential reports, for it can also be considered as dubitative (Travis, 2006) when expressing ‘speaker’s disbelief’ or, more generally, ‘‘lack of responsibility for the truth of the propositional content’’, as Olbertz (2007: 156) put it. This can be considered as an epistemic overtone that starts out as a pragmatic implicature then becoming increasingly semanticized as its syntactic scope narrows (Olbertz, 2007). Nonetheless, the intended meaning may be not restricted to a lack of responsibility: using dizque the speaker can also explicitly downgrade the epistemic commitment on the factuality of the reported situation. This is most probably the case in (11) and becomes more prominent in (12), a context that following Travis (2006) could be dubbed as ‘Labeling’, which corresponds to English ‘socalled’ in the translation proposed by Olbertz (2007: 163). The speaker’s disbelief is clearer in (13), where it is even possible that no speech act is reported: the subject may simply act as if he were ruling (‘pretending to rule’) without explicitly saying it. As Olbertz (2007: 161–162) observes, the absence of a speech act is particularly apparent in (14), where the speaker is the source of the information, which excludes reporting what a third person says. Contrary to its original meaning, the verb ‘say’ does not involve any actual speech act and here the speaker is only describing a non-factual situation (‘what he and his friends pretended to do and how they pretended to be’) (13) (14)
A los seis meses de andar dizque gobernando se puso enfermo. ‘After having gone about pretending to rule for six months he fell ill.’ [a group of boys is being shown a box of valuable essences] En eso andamos dizque maravillados cuando ¡chı´ngale! nos robamos la caja de las esencias y echamos a correr. ‘Meanwhile we go about supposedly amazed when, whoosh!, we steal the box with the essences and hit the road.’
As mentioned above, this meaning goes not only beyond the proper domain of evidentiality, crucially involving epistemicity, as Olbertz and Travis suggest, but it is also hardly captured as an epistemic overtone. What is expressed in (13–14), and most probably in (12) as well, is a totally pretended and therefore non-factual situation that is fictitiously created by the speaker and not only epistemically interpreted as dubious. Trying to reinterpret these data I propose the following semantic evolution for dizque: (15)
EVIDENTIALWEPISTEMIC-EVIDENTIALWNON-FACTUAL
Provided that ‘say’ is per se a neutral verb of report it can be assumed that the starting point of this evolution is ‘pure evidentiality’ without epistemic overtones. These overtones only arise in the second step of this evolution, the intermediate area dubbed ‘epistemic-evidential’, in which dizque is used as a reportive marker, being therefore evidential but also showing a reduction of
218
Mario Squartini
epistemic commitment, as may be the case in (11). Conversely, (13–14) represents a step forward in the semantic evolution, in which the form has totally dismissed its reportive meaning, being only a marker of reduction of factuality (‘non-factual’). In the spirit of this article, the basic point that can be made with respect to the semantic path sketched in (15) is connected to the role of non-factuality as the eventual step in the evolution of reportive evidentiality. This result can be capitalized with respect to our initial discussion on the semantic origin of the reportive meaning in the Romance Conditional. As shown above, synchronic data based on a comparison between French and Italian suggest that non-factuality may play a role as the semantic trigger of the process leading to the reportive meaning. Apparently, the evolution of dizque supports the hypothesis on the role of non-factuality, suggesting that dizque and the Conditional move along the same path even though displaying different orientations: while dizque moves toward non-factuality, the Conditional moves from non-factuality toward evidentiality. The semantic evolution of the Conditional can be represented in the following scheme, which reproduces (15) with the opposite orientation: (16) (EVIDENTIAL)oEPISTEMIC-EVIDENTIALoNON-FACTUAL The eventual step in (16), i.e., ‘(purely) evidential’ without any epistemic overtones, may have been already reached, but interpreting actual data in this sense is not uncontroversial and in fact different proposals can be found in the literature. Pietrandrea (2005: 103–104, 2007) considers the Italian reportive Conditional as purely evidential, opposing it to other epistemic-evidential forms (the modals dovere ‘must’ and potere ‘may, can’), while Kronning (2002a, b, 2005) emphasizes the mixed nature of the French reportive Conditional considering it as modal and evidential at the same time. Actually, the point whether the Romance reportive Conditional has reached or not the eventual stage dubbed ‘evidential’ can be left aside for the time being. The crucial point now is that the data under scrutiny confirm the diachronic connection between non-factuality and evidentiality. It is also interesting to observe that different markers, dizque and the Conditional, tend to share the same intermediate step, the one dubbed ‘epistemic-evidential’. The data presented in the next sections will show that the interpretation of other diachronic phenomena, derived from lexical and inflectional material, may provide a different picture, in which non-factuality does not show the expansive role toward evidentiality proposed in the analysis of the Romance Conditional, being rather a marginal feature that tends to be blocked or dismissed as the semantic evolution proceeds.
5. AVOIDING NON-FACTUALITY
IN
SEEM-VERBS
In Italian, as in many other languages, verbs roughly corresponding to English seem tend to develop epistemic as well evidential meanings. As Dixon (2005: 204) suggests, a copular verb
Evidentiality, epistemicity, and their diachronic connections to non-factuality
219
corresponding to English seem may signal that the speaker ‘‘is not fully certain whether the adjectival description is appropriate’’, which can be considered as epistemic, as is the case in the following Italian construction of the verb sembra: (17)
Carlo sembra alto ‘Carlo seems tall’
In addition, the speaker can also signal that his/her statement is based on indirect evidence. Without personally knowing Carlo I may utter (17) just looking at the length of his coat hanging on the hook. This is an inference based on indirect evidence, possibly an evidential usage, which can be found in different syntactic constructions of the same Italian verb sembrare (see Kratschmer, forthcoming). As shown in various languages,3 the semantic interpretation (inferential, but also reportive) of raising verbs such as seem is influenced by the syntactic construction in which they appear. It has been noted that the reportive meaning is particularly prominent in biclausal constructions with sentential complements such as (18). Here the Italian sembra has a reportive interpretation (‘those who have seen him say that he is tall’), also admitting the same inferential meaning as in (17). (18)
Sembra che Carlo sia alto ‘It seems that C. is tall’
Having sketched the basic synchronic distribution of modern Italian sembrare (see Kratschmer, forthcoming, for a much more detailed analysis) it is now worth considering its diachronic evolution. Being derived from a lexical basis attested in Latin the diachronic analysis can here include the etymology of the verb, which was not significant for the late Romance innovations discussed above (the inflectional Conditional and the lexicalized American Spanish dizque). Italian sembrare is a medieval loan from literary Provenc- al semblar, but if we trace back its semantic development from Old Occitan to Latin we find that the starting point of this diachronic process is connected to non-factuality. The etymological basis of the Provenc- al semblar (as well as French sembler) is late Latin similo, which derives from the adjective similis ‘similar’ and means ‘resemble’, but is also cognate to simulo ‘imitate, simulate’ (see also Barron, 1997, who elaborates on these etymological connections). Due to phonetic evolution the penultimate unstressed vowel has been lost, thus neutralizing the phonetic outcomes of similo and simulo, as is shown by French dissembler (Eng. dissemble), derived from Latin dissimulare ‘dissimulate’. Nonetheless, phonetic neutralization has not resulted in a semantic integration between the
3
See Cornillie (2004, 2007) on Spanish parecer, Kratschmer (forthcoming) on Italian pare/sembra, Dendale and Van Bogaert (2007) on French paraıˆtre/sembler and de Haan (2007) on Germanic SEEMverbs.
220
Mario Squartini
outcomes of simulo and similo, so that no semantic extensions expressing the idea of ‘imitation’ or ‘pretended resemblance’ can be found in the meanings of French sembler and Italian sembrare. This is most striking considering that these semantic features can be found in other lexical entries derivationally connected to sembler and sembrare. Consider Italian nouns sembiante ‘aspect, external appearance’ and sembianza ‘similarity, aspect, external appearance’, which are also loans from Provenc- al semblan and semblansa, both morphologically connected to the simplex semblar (the former is an original present participle of semblar and the latter a derived nominalization). If compared to the verb sembrare the Italian nouns sembiante and sembianza have a more descriptive meaning denoting external appearance, in particular connected to the human body, and what is most relevant, can also denote fake or pretended appearance or semblance (falso sembiante ‘fake appearance, semblance’). Conversely, the simplex verb sembrare tends to reduce its objective and descriptive meaning inherited from the original Latin ‘resemble’ and becomes more and more connected to a subjective and interpretative meaning in which speaker’s judgments play a crucial role. Thus, the notion of ‘resemblance’ (‘x resembles y’) has been gradually reinterpreted in a more subjective manner giving rise to the new meaning of ‘possible identity’ (‘x seems y’) or ‘possible adequacy to a given description’ (‘x seems to have the property y’), which is evaluated by the speaker and can therefore be intended as an epistemic judgment. Consistently with this tendency to reinterpret subjectively the notion of resemblance, syntactic constructions in which sembra behaves like the verb somigliare ‘resemble’ that used to be grammatical in Old Italian (19)4 have been lost nowadays. (19)
lo suo pensiero e lo suo senno no lo sappiamo divisare, non sembrava a quello dell’altre femine di niente. (Binduccio dello Scelto, La storia di Troia, 14th c.)5 ‘we cannot interpret her thought and intelligence, it was not at all similar to that of other women’
Conversely, constructions of sembra in which a pronominal clitic expresses the conceptualizer, who controls the epistemic judgment, flourish in modern Italian: (20)
Mi sembra che somigli a suo fratello ‘It seems to me that he looks like his brother’
The very fact that the verb somigliare ‘resemble’ can be construed in a complement clause dependent on the matrix verb sembrare ‘seem’ demonstrates that the original Latin meaning ‘resemble’ has been ‘bleached’ in sembrare. Lacking the original notion of resemblance its 4
Similar occurrences can also be found in Old French, as still documented by TLF, where the construction sembler a` qqn/qqc ‘be alike, be similar to someone/something’ is glossed as obsolete and considered synonymous of modern French ressembler a`. 5 The Old Italian data are extracted from the TLIO corpus.
Evidentiality, epistemicity, and their diachronic connections to non-factuality
221
extensions toward non-factual and imaginary worlds such as ‘pretended, simulated resemblance, or false identity’ become incongruous and therefore can only be found in derivational forms such as sembiante and sembianza. The diachronic evolution of Italian sembrare (and the same applies to French sembler as well, see footnote 4) demonstrates that a development into epistemic and evidential meanings may also prevent possible extensions toward non-factuality. Obviously this can be considered as contradicting evidence with respect to the evolution of the American Spanish dizque, in which non-factuality was interpreted as directly derived from evidentiality. Needless to say, this is also in contrast with the semantic evolution of the Conditional, where non-factuality was considered the trigger of the diachronic evolution toward evidentiality. Contrary to these cases, in the diachronic development of the verb sembrare non-factuality is apparently blocked by the evolution toward epistemicity and evidentiality. Note that the notion of non-factuality involved in the ‘seem’ predicates is comparable with the non-factual interpretations of the Conditional and dizque discussed above. As shown above, dizque can also refer to fictitious worlds such as the ‘pretended behavior’ in (14) and reference to possible and fictitious worlds is also the undisputable meaning of the Romance modal Conditional in (2) and (10). This makes the comparison of these results intrinsically contradictory: why is non-factuality lost in the semantic evolution of sembrare, while it coexists with evidentiality in the evolution of the Conditional and dizque? In what follows, additional evidence will be provided that can prove relevant in solving this inconsistency. In particular, another Romance diachronic phenomenon will be considered demonstrating that the loss of non-factual meaning, intended as the creation of fictitious worlds, is a more general process involving other epistemic forms.
6. OLD ITALIAN NON-FACTUAL FUTURE Like the synthetic future forms in other Romance languages, the Italian Future is not only temporally deictic, but it can also express the speaker’s own conjectures, irrespective of temporal reference. This conjectural Future is also documented in Old Italian (21), but the interesting point here is that the Old Italian Future could also occur in other modal contexts that would be barred nowadays (see Squartini, 2001 for details). A case in point is (22), where the Future refers to a fictitious world that the speaker does not desire. (21)
(22)
e ancora di cio` avrete saputo e ragionato con Bindo Squarcia e co. lLapo Chiari quando giunsero costa` (Lettere fiorentine, 13th c.) ‘and about that you must have learnt and discussed (lit. ‘will have learnt and discussed’) with B. S. and L. C. when they came there’ Scorseli la penna e scrisse CCC [ . . . ]. il Saladino parlo`: ‘‘Non dannare: scrivi CCCC. Per mala ventura se una tua penna sara` piu` larga di me!’’ (Novellino, 13th c.)
222
Mario Squartini
‘His pen slipped and wrote CCC [ . . . ]. Saladin said ‘‘Don’t correct it: write CCCC. It would be a misfortune, if a pen were (lit. ‘will be’) more generous than me!’’’ Elaborating on the modern ungrammaticality of (22), it can be concluded that the diachronic evolution of the modal Future tends to disentangle epistemic conjectural meaning based on the speaker’s subjective judgments (21) from pure non-factuality, i.e., reference to imaginary worlds (22). If compared to the provisional conclusions reached at the end of Section 4, these results confirm the observation that non-factuality can be incompatible with the diachronic evolution of forms having epistemic or evidential meanings such as the Future and the verb sembrare. With the former, non-factuality is so incompatible that, while possible in Old Italian, non-factual contexts have been diachronically dismissed. Conversely, the incompatibility with sembrare is demonstrated by the fact that extensions toward non-factuality (pretended and imaginary situations), though plausible and actually attested with etymologically connected words, are prevented with the verb sembrare itself. A possible explanation for this similar behavior can be connected to the common semantic functions of sembrare and the Future, particularly to their specialization in the inferential-conjectural domain. As opposed to the Conditional and the American Spanish dizque, which are consistently reportive markers,6 the Future and sembra are either only conjectural (Future) or both conjectural and reportive (sembra).7 Thus, the general point that can be made considering the whole set of data presented in this paper is that
6
It must be noted that this generalization is correct as far as the Italian Conditional is concerned while it cannot be extended as such to the French Conditional, which, unlike the Italian form, is also admitted in conjectural contexts: (i) On sonne; serait-ce Ge´raldine? (Melis, 2001: 75) ‘The bell rings; might it be G.?’ Consider, however, that direct questions are the only context in which the French Conditional can express a similar conjectural meaning. This suggests that the Conditional is compatible with epistemic conjectures only if associated with very low degrees of assertivity and factuality, as in direct questions, which strengthens the hypothesis of a positive correlation between non-factuality and the Conditional.
7
As for French sembler, the tendency to specialize as conjectural-inferential has been noted by Nølke (1994: 84–85). In comparing the quasi-synonymic pair of verbs paraıˆtre / sembler Nølke proposes the following glosses, which confirm the specialization of paraıˆtre as reportive as opposed to the intrinsic conjectural nature of sembler: (i) Il paraıˆt que Marie est malade ¼ On dit que Marie est malade ‘It seems that M. is sick ¼ It is said that M. is sick’ (ii) Il semble que Marie soit malade ¼ J’ai l’impression que Marie est malade ‘It seems that M. is sick ¼ I have the impression that M. is sick’
Actually, Dendale and Van Bogaert (2007: 76–79) observe that sembler can also have a reportive interpretation, but at the same time they consider the conjectural-inferential meaning as more salient than the reportive one, the latter being only contextually acquired.
Evidentiality, epistemicity, and their diachronic connections to non-factuality
223
non-factuality is positively connected to the evolution of reportive markers while it has a negative correlation with the evolution of conjectural-inferential markers. In the next concluding section this point will be further elaborated with respect to the more general relationship between epistemicity and evidentiality.
7. CONCLUSION As shown above, the contradictory behavior of the conjectural markers (Italian sembrare and the Future) as opposed to the Conditional and dizque seems to be consistently connected to their different functions (conjectural vs. reportive), with only reportive markers showing a positive correlation to non-factuality. Attempting to derive more general conclusions from these results, one should bear in mind, that while the evidential interpretation of the Romance Conditional and the American Spanish dizque is hardly disputable, the semantic nature of modal Futures and SEEMverbs is a thorny issue, a solution to which cannot be taken for granted. In most semantic analyses the Romance conjectural Future is interpreted as consistently epistemic, even by those who insist on the role of evidentiality in Romance (see Dendale, 2001 and Dendale and Van Bogaert, 2007: 79–83 on French and Pietrandrea, 2005 on Italian). Conversely, Rocci (2000: 249) considers the Italian Future as evidential (inferential), as well as Giacalone Ramat and Topadze (2007: 25–29) who subscribe to the evidential interpretation (with possible epistemic overtones) of both the Italian Future and the Italian ‘seem’ predicate sembrare. Given these inconsistencies, the diachronic data are difficult to interpret, but assuming that the Future and the conjectural uses of sembrare/sembler are (primarily) epistemic, our data seem to suggest that epistemicity is diachronically incompatible with non-factuality, while non-factuality turns out to be compatible with evidentiality as far as the Conditional and dizque are concerned. This result is surprising, if one considers that epistemicity expresses different degrees of commitment on the factuality of the situation, and might in principle be compatible with the endpoint in the gradient of factuality, i.e., non-factuality. Nonetheless, it must be borne in mind that epistemicity is normally intended as the expression of the speaker’s judgments (see Palmer, 1986), which can be based on external evidence (inferences) or general world knowledge (conjectures and assumptions). These judgments are not to be confused with the creation of possible or even counterfactual worlds, which are mental elaborations that do not need any connections to actual evidence or world knowledge. In a sense, this might mean that nonfactuality does not belong to the subjectification path, which instead turns out to be solely connected to subjective judgments on the actual world. It can be noted that this tentative conclusion is also supported by the bidirectional nature of the diachronic evolutions described above. As demonstrated by the comparative analyses of the Romance Conditional and the American Spanish dizque, non-factuality can be both a possible trigger of the diachronic
224
Mario Squartini
development toward reportivity (Conditional) as well as the eventual landing site of a reportive marker (dizque). Obviously, such a double orientation seems to be a problem in a model positing a unidirectional drive consistently fuelled by subjectification. Nonetheless, this ceases to be problematic if one assumes that non-factuality cannot be measured on a scale of subjectivity, the latter being only involved in subjective epistemic judgments about the actual world (inferences, conjectures, assumptions). However, leaving aside general issues connected to subjectification, the data presented above can be more specifically interpreted with respect to the directionality in the diachronic relationship between evidentiality and epistemicity. In this respect, these data confirm that reportive evidentials may develop an epistemic overtone (dizque) including non-factual meaning, while the only clear point with respect to the opposite direction is that non-factuality may evolve toward reportive evidentiality (Conditional). But, as has been said, non-factuality should not be confused with epistemic judgments, and nothing more general can be concluded with respect to the diachronic evolution of epistemicity, except for the tendency to avoid the coexistence of epistemic and nonfactual meanings. Thus, the possibility that ‘real’ epistemics evolve toward evidentials remains an open question that requires more extensive empirical coverage.
REFERENCES Aikhenvald, A. Y. (2004). Evidentiality. Oxford: Oxford University Press. Aikhenvald, A. Y. (2007). ‘‘Information source and evidentiality: What can we conclude?’’ in: M. Squartini (ed.), Special issue on Evidentiality between Lexicon and Grammar. Italian Journal of Linguistics 19: 209–227. Anderson, L. B. (1986). ‘‘Evidentials, paths of change, and mental maps: Typologically regular asymmetries’’, in W. Chafe and J. Nichols (eds.), Evidentiality: The linguistic coding of epistemology. Norwood, NJ: Ablex, 273–312. Barron, J. (1997). ‘‘LFG and the history of raising verbs’’, in M. Butt and T. H. King (eds.), Proceedings of the LFG97 Conference. San Diego, CA: University of California, CSLI. Caudal, P. and C. Vetters (2005). ‘‘Un traitement conjoint du conditionnel, du futur et de l’imparfait: Les temps comme des fonctions d’actes de langage’’, in A. Molendijk and C. Vet (eds.), Temporalite´ et attitude. Structuration du discours et expression de la modalite´, Cahiers Chronos 12. Amsterdam: Rodopi, 109–124. Cornillie, B. (2004). Evidentiality and epistemic modality in Spanish (semi-)auxiliaries. A functionalpragmatic and cognitive-linguistic account, Doctoral dissertation, Faculteit Letteren, Katholieke Universiteit Leuven. Cornillie, B. (2007). ‘‘The continuum between lexical and grammatical evidentiality: A functional analysis of Spanish parecer’’, in M. Squartini (ed.), Special Issue on Evidentiality between Lexicon and Grammar, Italian Journal of Linguistics 19: 109–128.
Evidentiality, epistemicity, and their diachronic connections to non-factuality
225
De Haan, F. (2001). ‘‘The relation between modality and evidentiality’’, in R. Mu¨ller and M. Reis (eds.), Modalita¨t und Modalverben im Deutschen (Linguistische Berichte, Sonderheft 9). Hamburg: Buske, 201–216. De Haan, F. (2005). ‘‘Coding of evidentiality’’, in M. Haspelmath, M. Dryer, D. Gil and B. Comrie (eds.), World atlas of language structures. Berlin: Mouton de Gruyter, 318–321. De Haan, F. (2007). ‘‘Raising as grammaticalization: The case of Germanic seem-verbs’’, in Squartini, M. (Ed.), Special issue on Evidentiality between Lexicon and Grammar, Italian Journal of Linguistics 19: 129–150. Dendale, P. (1993). ‘‘Le conditionnel de l’information incertaine: Marqueur modal ou marqueur e´videntiel?’’, in G. Hilty (ed.), Actes du XXe Congre`s international de linguistique et philologie romanes Vol. 1. Tu¨bingen/Basel: Francke, 163–176. Dendale, P. (1994). ‘‘Devoir e´piste´mique, marqueur modal ou e´videntiel?’’, in P. Dendale and L. Tasmowski (eds.), Les sources du savoir et leurs marques linguistiques. Langue franc- aise 102: 24–40. Dendale, P. (2001). ‘‘Le futur conjectural versus devoir e´piste´mique: Diffe´rences de valeur et de restrictions d’emploi’’. Le franc- ais moderne 69: 1–20. Dendale, P. and J. Van Bogaert (2007). ‘‘A semantic description of French lexical evidential markers and the classification of evidentials’’, in M. Squartini (ed.), Special issue on Evidentiality between Lexicon and Grammar, Italian Journal of Linguistics 19: 65–89. Dixon, R. M. W. (2005). A semantic approach to English grammar. Oxford: Oxford University Press. Giacalone Ramat, A. and M. Topadze (2007). ‘‘The coding of evidentiality: A comparative look at Georgian and Italian’’, in M. Squartini (ed.), Special issue on Evidentiality between Lexicon and Grammar, Italian Journal of Linguistics 19: 7–38. Korzen, H. and H. Nølke (2001). ‘‘Le conditionnel: niveaux de modalisation’’, in P. Dendale and L. Tasmowski (eds.), Le conditionnel en franc- ais (Universite´ de Metz, Recherches linguistiques 25). Paris: Klincksieck, 125–146. Kratschmer, A. (forthcoming). ‘‘Cate´gorisation vs comparaison: une question de quantification e´piste´mique. Mode`le interpre´tatif se´mantico-pragmatique modulaire des constructions italiennes avec sembrare/parere’’, in Cahiers Chronos 7. Colloque international sur la temporalite´ verbale, les aspects, les modes et la modalite´, Antwerp, 2006. Kronning, H. (2002). ‘‘Le conditionnel ‘journalistique’: Me´diation et modalisation e´piste´miques’’. Romansk Forum 16 (2): 561–575. Kronning, H. (2002). ‘‘Modalite´ et e´videntialite´’’, in M. Birkelund, G. Boysen and P. S. Kjærsgaard (eds.), Aspects de la modalite´. Actes du Colloque International sur la Modalite´ a` Odense 2000. Tu¨bingen: Niemeyer, 131–151. Kronning, H. (2005). ‘‘Polyphonie, me´diation et modalisation: Le cas du conditionnel e´piste´mique’’, in J. Bres, P. P. Haillet, S. Mellet, H. Nølke and L. Rosier (eds.), Dialogisme, polyphonie. Approches linguistiques. Bruxelles: De Boeck/Duculot, 297–312. Melis, L. (2001). ‘‘Hypothe`ses non temporelles sur le conditionnel comme tiroir de l’indicatif ’’, in P. Dendale and L. Tasmowski (eds.), Le conditionnel en franc- ais (Universite´ de Metz, Recherches linguistiques 25). Paris: Klincksieck, 67–88. Mortelmans, T. (2000). ‘‘On the ‘evidential’ nature of the ‘epistemic’ use of the German modals mu¨ssen and sollen’’, in J. van der Auwera and P. Dendale (eds.), Modal verbs in Germanic and Romance languages. Belgian Journal of Linguistics 14: 131–148. Nølke, H. (1994). ‘‘La dilution linguistique des responsabilite´s. Essai de description polyphonique des marqueurs e´videntiels il semble que et il para ˆıt que’’. Langue franc- aise 102: 84–94.
226
Mario Squartini
Olbertz, H. (2007). ‘‘Dizque in Mexican Spanish: The subjectification of reportative meaning’’, in M. Squartini (ed.), Special issue on Evidentiality between Lexicon and Grammar, Italian Journal of Linguistics 19: 151–172. Palmer, F. R. (1986). Mood and modality. Cambridge: Cambridge University Press. Pietrandrea, P. (2005). Epistemic modality. Functional properties and the Italian system. Amsterdam: John Benjamins. Pietrandrea, P. (2007). ‘‘The grammatical nature of some epistemic-evidential adverbs in spoken Italian’’, in M. Squartini (ed.), Special issue on Evidentiality between Lexicon and Grammar, Italian Journal of Linguistics 19: 39–63. Plungian, V. A. (2001). ‘‘The place of evidentiality within the universal grammatical space’’, in P. Dendale and L. Tasmowski (eds.), Special issue on Evidentiality, Journal of pragmatics 33: 349–357. Rocci, A. (2000). ‘‘L’interpre´tation e´piste´mique du futur en italien et en franc- ais: une analyse proce´durale’’, in J. Moeschler (ed.), Infe´rences directionnelles, repre´sentations mentales et subjectivite´. Cahiers de linguistique franc- aise 22: 241–274. Squartini, M. (1999). ‘‘Riferimento temporale, aspetto e modalita` nella diacronia del condizionale italiano’’. Vox Romanica 58: 57–82. Squartini, M. (2001). ‘‘Filogenesi e ontogenesi del Futuro italiano’’. Archivio glottologico italiano 86: 194–225. TLF ¼ Tre´sor de la langue franc- aise. Dictionnaire de la langue du XIXe et du XXe sie`cle. Paris: CNRS/ Gallimard. TLIO ¼ Tesoro della lingua italiana delle origini, Opera del Vocabolario Italiano, CNRwww.ovi.cnr.it Traugott, E. C. (1989). ‘‘On the rise of epistemic meanings in English: An example of subjectification in semantic change’’. Language 65: 31–55. Traugott, E. C. and R. Dasher (2002). Regularity in semantic change. Cambridge: Cambridge University Press. Travis, C. E. (2006). ‘‘Dizque: A Colombian evidentiality strategy’’. Linguistics 44: 1269–1297. Visconti, J. (2004). ‘‘Conditionals and subjectification. Implications for a theory of semantic change’’, in O. Fischer, M. Norde and H. Perridon (eds.), Up and down the cline. The nature of grammaticalization. Amsterdam: John Benjamins, 169–192. Ziegeler, D. (2003). ‘‘On the generic origins of modality in English’’, in D. Hart (ed.), English modality in context: Diachronic perspectives. Bern: Lang, 33–69.
Current Trends in Diachronic Semantics and Pragmatics Edited by Maj-Britt Mosegaard Hansen and Jacqueline Visconti r 2009 by Emerald Group Publishing Limited. All rights reserved.
12 THE GRAMMATICALIZATION OF NEGATIVE REINFORCERS IN OLD AND MIDDLE FRENCH: A DISCOURSE–FUNCTIONAL APPROACH Maj-Britt Mosegaard Hansen
1. INTRODUCTION In this paper, I propose a discourse–functional analysis of the uses of different forms of clause negation in Old French (9th–12th century) and Middle French (13th–16th century), namely the bipartite, ‘‘reinforced’’, forms ne . . . mie and ne . . . pas versus the simple preverbal negator ne, which at the time constituted the canonical form of clause negation in French. The literature has traditionally considered these three forms to have been synonymous, and hence in essentially random variation, in older stages of the French language (e.g., explicitly to this effect, Perle, 1878: 5; Sten, 1938: 30; Togeby, 1974: y258, and implicitly, Yvon, 1948; Price, 1962; Foulet, 1965; Harris, 1978; Winters, 1987).1 As will be shown below, however, a finegrained qualitative analysis of several Old and Middle French texts supports the hypothesis that there was, in fact, a functional differentiation between them for several centuries preceding the eventual generalization of ne . . . pas as the standard form.2 Indeed, a few previous studies
1
It should be noted that some of the above-mentioned authors do suggest that the choice between the forms may have been influenced by syntactic features or – in the case of poetry – metrical considerations, e.g., Perle (1878: 5) and Foulet (1965: 262). 2 There do not appear to be any differences in meaning and/or function between ne . . . mie and ne . . . pas in Old and Middle French texts. Ne . . . mie was gradually confined to certain non-dominant dialects before disappearing completely in the course of the 17th century, while ne . . . pas was preferred in other
228
Maj-Britt Mosegaard Hansen
(Guiraud, 1964; Martin, 1972; Schwegler, 1988: 41ff; Marchello-Nizia, 1999: 114) have already argued for the existence of a semantic and/or pragmatic difference. The present proposal, however, differs from theirs in salient ways (cf. the discussion in Section 3). Furthermore, I suggest an explanation of the subsequent grammaticalization of the bi-partite ne . . . pas as the standard negator of French, replacing the older preverbal ne, in terms of a very specific, and fairly frequent type of context in which reinforced negation is found to occur in Old and Middle French, and which I argue was the pivotal ‘‘bridging’’ context (Evans and Wilkins, 2000: 550; Heine, 2002: 84, redefined in Hansen, 2008: 62–63)3 that allowed for the functional reanalysis of the bipartite forms as expressing canonical negation. The structure of the paper is as follows: Section 2 discusses the overall diachronic evolution of French negation in terms of a series of stages known as ‘‘Jespersen’s Cycle’’; Section 3 presents the hypothesis pursued in this paper, and discusses the central notions of discourse status and markedness; in Section 4, I analyze a set of empirical data from a representative choice of Old and Middle French texts; finally, Section 5 is a conclusion.
2. FRENCH NEGATION
AND
JESPERSEN’S CYCLE
The diachronic evolution of standard negation in French is frequently cited as a textbook example of what, following Dahl (1979: 88), has become commonly known as Jespersen’s Cycle, whereby [t]he original negative adverb is first weakened, then found insufficient and therefore strengthened, generally through some additional word, and this in turn may be felt as a negative proper and may then in the course of time be subject to the same development as the original word (Jespersen, 1917: 4).
Thus, Old French used a simple preverbal ne (oLatin NON) as the standard marker of clause negation. At a very early stage, however, it became possible to reinforce this ne by a variety of
dialects, among which the one spoken in and around Paris. The remainder of this paper will not consider reasons for the eventual obsolescence of ne . . . mie, but some suggestions can be found in Hansen and Visconti (2009). 3 Heine (2002: 86) defines bridging contexts as contexts where, in addition to its ‘‘source’’ meaning, a given linguistic expression also allows for an innovative ‘‘target’’ meaning to be inferred. Whereas this author requires that the latter be the more plausibly intended of the two meanings, hence foregrounded with respect to the source meaning, Hansen (2008: 62f) argues that the innovative interpretation is backgrounded, and thus merely possible, in bridging contexts, and only comes into the foreground at the subsequent stage of evolution, namely in so-called ‘‘switch’’ contexts.
The grammaticalization of negative reinforcers in Old and Middle French
229
markers, most commonly pas (oLatin PASSU(M) ‘step’), mie (oLatin MICA(M) ‘crumb’), and point (oLatin PUNCTU(M) ‘point’), but also occasionally goutte (o Latin GUTTA(M) ‘drop’) and others besides (cf. Mo¨hren, 1980; Price, 1990). In the early stages, these markers can plausibly be assumed to have functioned as negative polarity items (or NPIs) (cf. Winters, 1987; Price, 1993; Detges and Waltereit, 2002: 176; Eckardt, 2003: Ch. 4). The most common ones, however, must very quickly have become grammaticalized as negative particles proper, as shown by the fact that they are found to co-occur, even in some of the earliest Old French texts, with nongrammaticalized NPIs: (1) (2)
Trestuz les alters ne pris jo mie un guant. (Roland, v. 3189) ‘All the others I don’t [mie] consider to be worth a glove.’ Tuit vo Franceis ne valent pas meaille. (Louis, v. 2433) ‘All your Frenchmen are not [pas] worth a dime.’
In the earliest extant French texts, we therefore already find variation, within one and the same text, between the plain negative ne and reinforced forms of negation, saliently ne . . . pas and ne . . . mie, cf. (1)–(2) above versus (3)–(4), from the same texts: (3)
(4)
Carles li reis, nostre emperere magnes, / Set anz tus pleins ad estet en Espaigne: / Tresqu’en la mer cunquist la tere altaigne. / N’i ad castel ki devant lui remaigne; (Roland, vv. 1–4) ‘Charles, the King, our great emperor, / Has been in Spain for a full seven years: / All the way to the sea did he conquer the high country. / There is not a castle that resists him;’ Li mieldre reis ot a nom Charlemaine; / Cil aleva volentiers dolce France; / Deus ne fist terre qui envers lui n’apende; (Louis, vv. 14–16) ‘The greatest king bore the name Charlemagne; / He happily made sweet France greater; / God made no country that doesn’t belong to him;’
Indeed, this seemingly random variation persists for several centuries, until ne . . . pas is finally fully grammaticalized as the canonical way to express clause negation by the end of the 17th century. The cycle has subsequently gone one step further in as much as speakers overwhelmingly tend to drop the preverbal ne in contemporary spoken French of the less formal variety, clause negation being thus expressed solely by the original postverbal reinforcer, pas (e.g., Ashby, 1981, 2001; Coveney, 1996, Ch. 3; Martineau and Mougeon, 2003; Hansen and Malderez, 2004; FonsecaGreber, 2007), as in (5): (5)
enfin c[Ø]’est pas la meme race c[Ø]’est pas la meˆme population (Voyage en Egypte, p. 33) ‘anyway it’s not the same race it’s not the same population’
230
Maj-Britt Mosegaard Hansen
Table 12.1. The evolution of French clause negation (sample sentence: ‘‘I do not say . . . ’’) Stage 1.
je ne dis
The negator is preverbal
Stage 2.
je ne dis (pas)
The negator is optionally strengthened by a postverbal element
Stage 3.
je ne dis pas
The strengthener grammaticalizes as part of a discontinuous negator embracing the verb
Stage 4.
je (ne) dis pas
The original negator becomes optional
Stage 5.
je dis pas
The negator is postverbal
Source: Adapted from Van der Auwera and Neuckermans (2004: 459).
Real-time studies by Ashby (1981, 2001) and dialectal data from Canadian and Swiss French (Martineau and Mougeon, 2003; Fonseca-Greber, 2007) suggest that ne is not at all unlikely to eventually disappear, at least from the spoken language, such that we may operate with the five stages of diachronic development seen in Table 12.1. Various hypotheses have been put forward to explain the changes involved in these five stages. Here, I will concentrate on Stages 1–3, as the ones that are relevant to this paper. As seen above, Jespersen (1917) proposes that the change from Stage 1 to Stage 2 can be explained largely by the phonetic weakening of the Latin negator NON to ne /n /, creating an increased need for reinforcing expressions. There are problems with this hypothesis, however: First of all, it does not explain why ne nevertheless remained perfectly capable of negating clauses without the help of a reinforcing element for centuries following the sound change in question. Secondly, according to Martineau and Mougeon (2003: 123–124), ne actually only became unstressed at the end of the Middle-French period, and the schwa-deletion that is now common in spoken French (as in Je n’sais pas, ‘I don’t know’) did not set in prior to that time, either. Thirdly, other Romance languages in which Latin NON was either not phonetically reduced at all (such as Italian) or was not reduced to the same extent as in French (such as Catalan) have nevertheless developed reinforced, bi-partite, forms of negation as alternatives to the simple preverbal no(n), as shown by the contrasting Catalan examples in (6)–(7) (see Section 3). Finally, reinforced negation is found already in pre-Classical Latin, in texts having an oral and/or colloquial tenor, e.g., the comedies of Plautus (cf. (8)). Indeed, the origins of Old French ne . . . mie (as well as of Old and Modern Italian non . . . mica) can in all probability be traced back to 1st century AD uses of Latin MICAM in contexts like (9): e
(6) (7)
L. Wittgenstein no va ser un lingu¨ista. L. Wittgenstein no va ser pas un lingu¨ista. (both from Espinal, 1993: 354, her (1a)-(1b)) ‘L. Wittgenstein was not [pas] a linguist.’
The grammaticalization of negative reinforcers in Old and Middle French
(8)
(9)
231
AM. Haec sacerdos Veneris hinc me petere aquam iussit a vobis. SC. At ego basilicus sum: quem nisi oras, guttam non feres. (Plautus, Rudens, 3rd–2nd cent. BC – from LLT) ‘AM. That priestess of Venus ordered me here to get some water from you. SC. But I’m an important person: if you don’t entreat me, you’ll not take a drop away.’ quinque dies aquam in os suum non coniecit, non micam panis (Petronius, Satyricon, 1st cent. AD – from LLT) ‘for five days he didn’t put any water in his mouth, not a crumb of bread’
Based on the word order theory of Lehmann (e.g., 1973), both Vennemann (1974: 366ff ) and Harris (1978: 23ff) suggest that typological pressures were responsible for the shift from preverbal to embracing – and increasingly just postverbal – negation, but they agree neither on the categorial status of the negators nor on the basic typological status of the different stages of French. Thus, according to Vennemann, the negating particles are adverbs, and French was developing from the SXV type, which in its consistent realization has adverbs before the verb, to the SVX type, which in its consistent realization has adverbs after the verb (Vennemann, 1974: 367).
For Harris, on the other hand, as indeed for Lehmann (1973), negation is a sentential operator. As such, it should therefore, according to Lehmann’s theory, occur after the verb in an OV language (i.e., what Vennemann refers to as a SXV language in the above quotation), while preceding it in a VO language. In fact, Harris explains the negative cycle in French as prompted by a typological shift in a completely different direction from that hypothesized by Vennemann, interpreting Old French as an SVO language, and contemporary spoken French as a VSX language (Harris, 1978: 118), in which the subject and object clitics have become prefixes on the verb, forcing ne out (the latter suggestion is echoed by Martineau and Mougeon (2003: 139ff )). This hypothesis, of course, raises the question of why pas has not then become preverbal in contemporary spoken French. In any case, given that the data lend themselves to such different interpretations at the syntactic level, it seems worthwhile to explore the alternative possibility that the rise of negative reinforcers may originally have been prompted by semantic/pragmatic considerations, rather than by purely syntactic pressures. In a relatively recent paper, Detges and Waltereit (2002: 176ff ) propose that at Stage 2, the use of ne . . . pas/mie etc. expressed that the state-of-affairs being negated was not realized to even the smallest degree; in other words, ne . . . pas/mie etc. would, at that stage, have been equivalent to modern English not at all. According to these authors, the change from Stage 2 to Stage 3 came about as a result of increasingly frequent unwarranted uses of this originally expressive discourse strategy, which allowed speakers to make their negative statements appear more informative, hence more interesting to the hearer. While this form of ‘‘rhetorical overuse’’ of the reinforced
232
Maj-Britt Mosegaard Hansen
construction may well have played a role in the grammaticalization of ne . . . pas, it does not, however, appear to be the whole story, as it leaves the actual pattern of variation found in Old and Middle French texts largely unexplained. For instance, it is not at all intuitively clear why the 13th-century hagiographer Jean de Joinville should have been concerned to make the underlined negative statement in (10), but not that in (11), appear particularly informative: (10)
(11)
Le non de ceulz qui estoient chevaliers entour le roy sont tiex : mon seigneur Geffroy de Sargines, mon seigneur Mahi de Marley, mon seigneur Phelippe de Nanteul, mon seigneur Hymbert de Biaujeu, connestable de France, qui n’estoit pas la, ainc- ois estoit au dehors de l’ost, . . . (Joinville, P173) ‘The names of those knights who were in the King’s entourage are the following: My Lord Geoffroy de Sargines, My Lord Mathieu de Marley, My Lord Philippe de Nanteuil, My Lord Imbert de Beaujeu, constable of France, who wasn’t [pas] there, but who was outside the camp, . . . ’ Li chevaliers ne fu pas esbahiz, ainc- ois le prist par la chape, et li dist que il ne le lairoit jusques a` tant que il averoit finei a` li. (Joinville, P91) ‘The knight was not embarrassed, but took him by the cloak, and told him that he would not let him go until he had a ransom from him.’
The contrast between (10) and (11) can, however, be straightforwardly explained in terms of the different discourse statuses of the two statements in question.4 It is now time to develop the central hypothesis of this paper.
3. HYPOTHESIS
OF THE
PRESENT PAPER
The hypothesis defended in this paper is that the use of ne . . . mie/pas in Old and Middle French, i.e., at Stage 2 of Jespersen’s Cycle as laid out in Table 12.1, was governed by discourse– functional constraints, which were gradually loosened, plausibly due to the frequent occurrence of these forms of negation in a very specific type of context, to be referred to below as ‘‘Janus-faced’’ contexts, thus allowing the transition to Stage 3 of the Cycle, viz. the obligatorification of the postverbal negator. Initial empirical support for this hypothesis was adduced in Hansen and Visconti (2009), and the present paper strengthens the case for its correctness by explicitly contrasting the uses of ne . . . mie/pas and plain ne in texts from the relevant period.
4 Anticipating a little, the affirmative proposition underlying the pas-marked relative clause in (10) represents information that is highly inferrable from the immediately preceding text. (The same can be said, by the way, of the affirmative proposition underlying the pas-marked clause in (11).) The plain ne-marked clause in (11), on the other hand, expresses information that is new to the hearer/reader.
The grammaticalization of negative reinforcers in Old and Middle French
233
I take my point of departure in observations that have been made in the literature about the pragmatics of reinforced negative forms in certain contemporary Romance vernaculars which, unlike Modern Standard French, canonically express clause negation by a single preverbal negative particle, namely Italian, Catalan, and Brazilian Portuguese. 3.1. Discourse constraints on non-canonical negation in other Romance languages Several studies have linked the use of Italian non . . . mica (a cognate of Old French ne . . . mie), Catalan no . . . pas (a cognate of French ne . . . pas), and Brazilian nao˜ . . . nao˜ to the notion of presupposition (e.g., Schwegler, 1988; Espinal, 1993; Bernini and Ramat, 1996; Zanuttini, 1997), suggesting that the reinforcing markers in question are restricted to occur in contexts where the proposition being denied is presupposed or otherwise part of the common ground. In fact, Schwegler (1988: 41ff) suggests that the use of Old and Middle French ne . . . mie/pas was subject to a similar constraint.5 According to Zanuttini (1997: 61), for instance, in Italian only (13) below would constitute a felicitous exchange with the addition of mica, whereas (12) would not: (12)
(13)
A. Chi viene a prenderti? B. Non so. Ma Gianni non a (*mica) la macchina. ‘A. Who’s coming to pick you up? B. I don’t know. But Gianni doesn’t [mica] have the car.’ A. Chi viene a prenderti – Gianni? B. Non so. Ma Gianni non a mica la macchina. ‘A. Who’s coming to pick you up – Gianni? B. I don’t know. But Gianni doesn’t [mica] have the car.’
While not rejecting these proposals, Schwenter (2006) observes that they are not quite accurate as they stand. For one thing, the notions of presupposition and common ground are not clearly defined in any of the studies cited. Secondly, prior belief in the corresponding affirmative proposition is not a necessary condition for the felicitous use of these reinforced forms of negation. Catalan no . . . pas and Italian non . . . mica, for instance, may be found in confirmatory contexts like (14): (14)
5
A. La Maria non viene a quest’ora. B. Effettivamente, la Maria non viene mica a quest’ora.
In a not unrelated vein, Guiraud (1964), Martin (1972), and Marchello-Nizia (1999: 114) analyze the difference between plain ne and ne . . . mie/pas in semantic terms, as involving the virtuality versus actuality of the proposition. These notions, which are crucial to their accounts, remain, however, undefined, hence rather too vague to be really useful. Furthermore, none of these authors make any attempt to explain what led to the meaning change that eventually allowed ne . . . pas to become generalized as the canonical negator in Modern French.
234
Maj-Britt Mosegaard Hansen
‘A. Maria won’t be coming this late. B. True, Maria won’t [mica] be coming this late.’ Based on examples from Brazilian Portuguese similar to (12)–(14), Schwenter instead proposes two constraints on the discourse status of ‘‘non-canonically negated’’ propositions: (1) that proposition or its underlying positive variant must be ‘‘discourse-old’’, as defined by Prince (1992), and (2) it must be contextually activated in the sense of Dryer (1996). These notions will be discussed further in the immediately following section. However, while I am not questioning the correctness of Schwenter’s proposal as far as Brazilian Portuguese is concerned, and while it does account for many of the Old and Middle French examples in my database, as well, it appears, nevertheless, to be in need of some refinement in order to be able to account for all of them. In particular, the definition of ‘‘discourse-old’’ propositions will be amended. At the same time, the idea that non-canonically negated propositions must be fully activated has to be nuanced in the case of Old and Middle French. 3.2. The notion of discourse status Prince (1992) classifies discourse entities along two intersecting parameters, viz. ‘‘discourseold’’ versus ‘‘discourse-new’’ and ‘‘hearer-old’’ versus ‘‘hearer-new’’, as in Table 12.2. Discourseold entities are defined as having been explicitly evoked in prior discourse; such entities are therefore also hearer-old. Discourse-new entities, on the other hand, may be either brand-new to the hearer or part of his6 encyclopedic knowledge store. There is, however, a problem with Prince’s taxonomy, namely that the status of inferrable information remains unclear, a fact noted by Birner (2006: 17), who refines the model by defining discourse-old information in terms of inferential links rather than prior mention. Inferential links can work in a ‘‘forward’’ manner, being automatically invited by a given trigger, such as the link from the trigger exams to grades in (15) (this type of inference is also referred to in the literature Table 12.2. Prince’s (1992) taxonomy of given/new information Hearer-old
Hearer-new
Discourse-old
Previously evoked
N/A
Discourse-new
Not evoked but known
Brand-new
6 In this paper, speakers will be anaphorically referred to using the feminine pronoun, while hearers will be referred to as masculine, except where authentic (i.e., non-constructed) examples that make this gender distribution awkward are being commented on.
The grammaticalization of negative reinforcers in Old and Middle French
235
as ‘‘elaborating inferences’’). Alternatively, they can be ‘‘backward’’, or ‘‘bridging’’, inferences, like the link from vacation to in Spain to day trip to Gibraltar in (16); i.e., hearers will not automatically infer that a vacation in Spain entails a day trip to Gibraltar, but once mention is made of such a day trip, they will typically establish textual coherence by assuming such a connection: (15) Jane passed all her exams this semester. Some of her grades were not terrific, though. (16) We had a good vacation in Spain. The day trip to Gibraltar was fun. As elaborating inferences may be inferences of identity, information that has been explicitly mentioned in the previous discourse is accommodated by this category, while the existence of bridging inferences allows Birner (2006: 25) to fill in the missing cell representing the intersection of discourse-old and hearer-new in Table 12.2, as seen in Table 12.3.
Table 12.3. Birner’s revised taxonomy of given/new information Hearer-old
Hearer-new
Discourse-old
Evoked: Identity/Elaborating Inferrable (inferentially linked and known to hearer)
Bridging Inferrable (inferentially linked but not known to hearer)
Discourse-new
Unused (not inferentially linked but known to hearer)
Brand-new (neither inferentially linked nor known to hearer)
This paper will retain Birner’s model of information status, whereby discourse-old information may be not only previously evoked, but also backwards inferrable. As will be seen below, that model provides a better account of some uses of Old and Middle French ne . . . mie/pas than does Prince’s original version of the taxonomy. When the notion of givenness versus newness in discourse is extended to cover propositions, as it must be when the object of study is clause negation, the issue of belief is raised. In the model suggested in Schwenter (2006), this is where Dryer’s (1996) notion of activation comes in. For Dryer, a proposition can be given in two distinct senses: it may be ‘‘activated’’, i.e., present to the attention of the hearer at a given stage of the discourse, or it may be ‘‘presupposed’’, i.e., part of the common ground, the set of propositions that the speaker believes and assumes the hearer to believe, as in Stalnaker’s (1991[1974]: 473) definition. Clearly, the two are independent – although of course not mutually exclusive – properties. In view of examples such as (14) above, what is important for the (non-)use of reinforced negators must be activation of, not belief in, a given proposition. As already adumbrated, however, some of my Old and Middle French
236
Maj-Britt Mosegaard Hansen
examples of reinforced negation pose problems for a requirement that the proposition be fully activated in the hearer’s mind at the time of utterance, in as much as they represent bridging inferrables (see Section 4.3.3). It seems therefore, that to accommodate this type of example, Schwenter’s activation constraint must be weakened to a requirement that the proposition be accessible on the basis of already activated information. Such a weakening is, of course, fully in line with Birner’s (2006) revised definition of ‘‘discourse-old’’ entities. 3.3. The notion of markedness and the Old/Middle French negators The above discussion, both of Jespersen’s Cycle and of the suggested discourse-functional constraints on the use of reinforced negation in Old and Middle French, raises a further issue, namely that of markedness. Following the definition of Givo´n (1990: 945), three types of markedness, which can, but need not, coincide, must be distinguished in language: (1) Structural markedness, whereby the marked item has a larger and/or a more complex structure than the unmarked item. In most languages, for instance, the plural of nouns will be marked with respect to the singular, e.g., dog (unmarked) versus dogs (marked). (2) Cognitive markedness, whereby the semantic range of the marked item constitutes a subpart of the range of the unmarked item. Thus, for instance, the noun bitch is semantically marked with respect to the noun dog in English, as the latter can refer to both male and female animals. (3) Textual markedness, whereby the unmarked item will be statistically more frequent across text types than the marked item. The qualifier ‘‘across text types’’ is important, in as much as certain types of texts may, of course, have more frequent recourse to the marked term (for instance, a text about diseases specific to female animals, as opposed to male animals, might use the word bitch more frequently than the word dog, even though the overall frequency of the latter in English-language texts is no doubt the greater). It is clear that the plain preverbal ne was the unmarked item and reinforced negation the marked item in Old French, according to all three criteria. It is equally clear that in Modern French, the bi-partite, ‘‘embracing’’, negation is cognitively and textually unmarked, the plain ne having become confined to a handful of syntactically definable contexts and to largely formal registers. Furthermore, even in these contexts, the choice of the plain form is a mere option, the bi-partite negation being always an acceptable alternative. In Modern ‘‘standard’’ French, the bipartite negation remains, of course, structurally marked, although the almost ubiquitous dropping of ne in informal spoken French puts the two competing negators on an equal footing in that respect.
The grammaticalization of negative reinforcers in Old and Middle French
237
In other words, a gradual unmarking of the reinforced form of negation, and a concomitant ‘‘marking’’ of the plain form, must have taken place in the course of Middle French.
4. THE USE OF ‘‘PLAIN’’ AND MIDDLE FRENCH
VERSUS
‘‘REINFORCED’’ NEGATION
IN
OLD
In order to substantiate the hypothesis outlined above, it is now time to proceed with an analysis of textual data from older stages of French.
4.1. Data The data of the present study are drawn from the four texts below, representing a period of just over two centuries. These texts were chosen from among a total of 74 available from the searchable electronic corpus Base de franc- ais me´die´val (henceforth, BFM): La chanson de Roland (approx. 1090) Le couronnement de Louis (approx. 1130) La queste del Saint Graal (approx. 1220) Jean de Joinville: Histoire de Saint Louis (1298–1309) The former two are epic poems, while the latter two are prose texts, respectively, a romance and a hagiography. Using poetry as data is not ideal because considerations of meter and rhyme may conceivably have entered into the choice of a particular form of negation; however, the nature of the extant Old French texts is such that a large enough database consisting wholly of prose was simply not available. In Roland and Louis, there are still relatively few tokens of reinforced negation, and it was therefore possible to analyze all of them. The number of tokens rises very steeply in the Graal, however, and, in the case of ne . . . pas, stays quite high in Joinville’s text. In those latter cases, I have therefore submitted only the first 50 or so tokens in those texts to closer scrutiny. The data were found by searching the BFM for tokens of mie and pas, respectively, in those texts. As for the simple preverbal negation ne, I have analyzed the first 50 or so examples from each of the texts. In order, on the one hand, to avoid ‘‘noise’’ in the form of tokens of ne as a coordinating conjunction or in combination with reinforcing forms, including markers of constituent negation (e.g., ne . . . personne), and, on the other hand, to include tokens of ne written together with an object pronoun (e.g., nel), these tokens had to be culled manually, from printed editions of the texts in question. For that reason, no data on the total number of tokens of plain ne in the four texts are provided. The figures are then as in Tables 12.4–12.6.
238
Maj-Britt Mosegaard Hansen
Table 12.4. Number of tokens of mie analyzed in each individual text Roland
Louis
Graal
Joinville
44
19
50/228a
8
a
Numbers following a slash (/) represent the total number of tokens in the text in question as a whole (according to the BFM).
Table 12.5. Number of tokens of pas analyzed in each individual text Roland
Louis
Graal
Joinville
4
15
50/200
53/143
Table 12.6. Number of tokens of plain preverbal ne analyzed in each individual text Roland
Louis
Graal
Joinville
50
50
51
50
4.2. Syntactic status of the negated clauses My data reveal one very clear trend in the syntax of the three competing forms of negation, namely that, over time, all three – but the plain ne in particular – increasingly come to be used in subordinate clauses, as opposed to independent or main clauses, cf. Table 12.7 (examples are given in (17)–(22)): Independent clauses: (17) (18) (19)
Jo nen ai ost qui bataille li dunne, (Roland, v. 18) ‘I don’t have an army that can combat him,’ Dist Cları¨en: ) Dame, ne parlez mie itant! * (Roland, v. 2724) ‘Clarien said, ‘‘My Lady, don’t talk [mie] so much !’’’ Tuit vos Franceis ne valent pas meaille. (Louis, v. 2433) ‘All your Frenchmen are not [pas] worth a dime.’
Subordinate clauses: (20)
Deus ne fist terre qui envers lui n’apende; (Louis, v. 16) ‘God made no country that doesn’t belong to him ;’
The grammaticalization of negative reinforcers in Old and Middle French
239
Table 12.7. Distribution of the negators according to clauses status Roland (%)
Louis (%)
Graal (%)
Joinville (%)
Indep. cl.
86.3
94.7
70
37.5
Subord. cl.
13.6
5.2
30
62.5
Ne . . . mie
Ne . . . pas Indep. cl.
75
93.3
64
58.4
Subord. cl.
25
6.6
36
41.5
Plain ne Indep. cl.
80
66
32
16
Subord. cl.
20
34
68
84
(21)
(22)
Quant il c- o vit que n’en pout mie freindre, A sei meı¨sme la cumencet a pleindre: (Roland, vv. 2314–15) ‘When he saw that he could not break [mie] it, he started to complain to himself :’ Ja mosterront qu’il ne sont pas ami. (Louis, v. 2562) ‘Soon they’ll show that they are not [pas] friends.’
While this distribution probably in part reflects a generic characteristic of medieval French epic poetry (or chanson de geste), where parataxis plays a very important role, this diachronic pattern is nevertheless to be expected in the case of the reinforced forms, ne . . . mie/pas. Both are grammaticalizing forms, and one of the characteristics of such items is held to be that they typically occur first in independent clauses, which tend to be more innovative (Givo´n, 1976: 170). In other words, the trend observed in Table 12.7 can plausibly be explained as a gradual broadening of the syntactic potential of ne . . . mie/pas. Given that the plain preverbal ne is inherited from the mother language, Latin, as the canonical form of negation, such a scenario is unlikely to be applicable in explaining why that form comes to be almost exclusively used in subordinate clauses over the centuries. In this case, where we are dealing with a form that is gradually being ousted, it is probably more appropriate to speak of an increasing confinement to subordinate clause types, as these tend to be more syntactically conservative than independent clauses (cf. Bybee et al., 1994: 296). Both trends can thus be explained as part and parcel of the shift in markedness discussed in Section 3.3.
240
Maj-Britt Mosegaard Hansen
For further observations about the syntactic features of the contexts in which ne . . . mie/pas appear, but which do not appear to be decisive, the reader is referred to Hansen and Visconti (2009: 144ff). Additional comments on the syntax of the simple preverbal negator will be made below, in connection with the semantic/pragmatic analysis of the examples. 4.3. Discourse contexts of reinforced versus plain negators The following sections present a detailed analysis of the contexts in which clause negation using either ne . . . mie/pas or plain ne is found in my data. I will start by considering the contexts of the two reinforced negators, and subsequently compare them to the contexts in which the plain ne occurs. 4.3.1. Ne . . . mie/pas. In terms of their discourse–functional status, the examples of reinforced negation in my database can be subclassified according to the perceived degree of givenness of either the negative proposition itself or its underlying positive proposition. The two types of cases appear to be fairly evenly distributed across the four texts, supporting the observation made by Schwenter (2006), already mentioned in Section 3.1, according to which prior activation of the corresponding affirmative is not a constraint on the use of reinforced negation in Romance. Four broad classes of examples can be discerned, as follows: a. Examples where the ne . . . mie/pas-marked clause represents a denial or rejection of part of the preceding text; b. Examples where the ne . . . mie/pas-marked clause represents a repetition or paraphrase of part of the preceding text; c. Examples where the ne . . . mie/pas-marked clause represents either the expression or the denial/rejection of a (pragmatic) presupposition; d. Examples where the ne . . . mie/pas-marked clause represents either the expression or the denial/rejection of another type of inference warranted by the previous text. The four categories will be discussed in greater depth below.7 We will see that the borders between them may be fuzzy, in that some of the examples falling within in categories (a)–(c) may nevertheless involve some degree of inference. 4.3.1.1. Denial/rejection of part of the preceding text. The data contain a number of straightforward examples of direct denial or rejection of a preceding utterance. This category accounts for between 5.2% and 25% of the examples of ne . . . mie in the individual texts. As for
7
Readers who wish to see relevant examples of ne . . . mie/pas from all four texts in the database are referred to Hansen and Visconti (2009).
The grammaticalization of negative reinforcers in Old and Middle French
241
ne . . . pas, 20% of the examples in La queste del Saint Graal and 20.7% in Joinville belong here. No instances were found in Le coronemenz de Loois, and only one example (namely (23) below) in La chanson de Roland. As there are only altogether four tokens of ne . . . pas in the latter text, I have chosen not to calculate a percentage in this case. (23)
(24)
Fols est li reis ki vos laissat as porz. [ . . . ] ) Ultre, culvert ! Carles n’est mie fol, . . . * (Roland, vv. 1193, 1207) ‘Mad is the king who left you in these passes. [ . . . ] ‘‘Out of my sight, villain! Charles is not [mie] mad, . . . ’’’ ) Dame, soffrez que nostre noviaus chevaliers viegne avec nos a la cort mon seignor le roi. [ . . . ] – ) Sire, fet ele, il n’ira pas ore ; . . . * (Graal, p. 3) ‘‘‘My Lady, allow our new knight to come with us to the court of our Lord the King.’’ [ . . . ] – ‘‘Sir, says she, he’ll not [pas] go now; . . . ’’’
Some examples in this category, however, involve a more indirect form of denial or rejection, where an inference is needed for the mie/pas-marked clause to be understood in this way. Thus, in (25), the reader knows that the first speaker is the villain of the story, and that the proposal he makes to the second speaker, the hero, implies that the latter should betray his king: (25)
) . . . Faisons la paiz et seions bon ami, Et je et tu avrons Rome a tenir. * [ . . . ] ) . . . Je ne vueil mie mon dreit seignor traı¨r, . . . * (Louis, vv. 2531–2, 2535) ‘‘‘Let’s make peace and be good friends, And you and I will hold Rome.’’ [ . . . ] ‘‘ . . . I’ll not [mie] betray my rightful lord, . . . ’’’
4.3.1.2. Repetition or paraphrase. This category comprises 12% and 38.6% of the examples of ne . . . mie in the three older texts, and between 11.3% and 20% of the clauses containing ne . . . pas in the three later text. I have found no examples of this kind with ne . . . mie in Joinville, and none with ne . . . pas in La chanson de Roland: (26)
(27)
S’il le redote, nuls n’en deit merveillier. [ . . . ] S’il le redote, ne fait mie a blasmer. (Louis, vv. 675, 686) ‘If he fears him, no-one should wonder. [ . . . ] If he fears him, he’s not [mie] to blame.’ . . . de dous mille et huit cens chevaliers que li roys mena en Egypte, ne l’en demoura que sept cens . . . [ . . . ] . . . pour ce que il ne li estoit pas demoure´ la tierce partie de ses gens; (Joinville, P147, 149) ‘ . . . out of two thousand and eight hundred knights that the King brought with him to Egypt, only seven hundred remained . . . [ . . . ] . . . because not [pas] even a third of his men had remained with him ;’
As was the case with denial/rejection, some examples in this category seem to require some degree of inference to be understood as repetition or paraphrase; for instance, (28), where the
242
Maj-Britt Mosegaard Hansen
general statement made in the ne . . . mie-marked clause paraphrases a particularized statement in the preceding text: (28)
) Belz fils *, dist il, ) envers mei entendez: Vei la corone qui est desus l’altel: par tel covent la te veuil je doner: Tort ne luxure ne pe´chie´ ne mener, [ . . . ] Filz Looı¨s, a celer ne te quier, Quant Deus fist reis por peuple justicier, Il nel fist mie por false lei jugier, faire luxure, ne alever pechie´, . . . (Louis, vv. 62–65, 174–177) ‘‘‘Dear Son’’, says he, ‘‘listen to me : See the crown that is on the altar: on such a condition will I give it to you: that you commit neither injustice, nor debauchery, nor sin, [ . . . ] Louis, my son, I do not wish to conceal it from you, when God made kings to dispense justice to people, he did not [mie] do it so they could pronounce unjust sentences, commit debauchery, nor exalt sin, . . . ’
4.3.1.3. Expression or denial of a (pragmatic) presupposition. This third category of examples accounts for between 5.2% and 37.5% of the examples of ne . . . mie in the four texts, and for between 11.1% and 26.6% of the examples of ne . . . pas in the three later texts. As with denial/ restriction, one out of a total of four examples of ne . . . pas in La chanson de Roland fits this category, but, as before, stating a percentage on such a small number of tokens would be misleading: (29)
(30)
) Sire, coment avez vos non, . . . *-) De mon nom, fet il, ne puez tu mie savoir : . . . (Graal, p. 29) ‘‘‘Sir, what is your name, . . . ’’ – ‘‘Of my name, says he, you cannot know [mie]; . . . ’’’ Et je li respondi que je n’en avoie pooir; [ . . . ] Le le´gat se courouc- a a moy et me dit que je ne le deusse pas avoir refuse´. (Joinville, P420-21) ‘And I replied that I didn’t have the power to do so ; [ . . . ] The legate became angry with me and told me that I should not [pas] have refused.’
It should be noted that when ne . . . mie/pas is used in direct speech, as in (31), the pragmatic presupposition in question may be part of the perceptual common ground. It may also be part of general cultural knowledge, as in (32): (31)
(32)
) Oncles Guillelmes *, dist Bertrans li guerriers, ) De vostre brant vei sanglant tot l’acier, Et vostre escuz n’est mie toz entiers: . . . (Louis, vv. 2190–92) ‘‘‘Uncle Guillaume’’, says Bertrand the warrior, ‘‘I see the steel of your sword is all bloody, And your shield is not [mie] quite intact: . . . ’ ) Dreiz emperere, entendez mon langage ; Ne vos salu, n’est pas dreiz que le face . . . (Louis, vv. 2388–89) ‘‘‘Just emperor, hear my words; I don’t salute you, it’s not [pas] appropriate that I should . . . ’
4.3.1.4. Expression or denial of an inference warranted by the preceding text. This fourth and last category is the numerically most important one. Thus, between 37.5% and 58% of the clauses
The grammaticalization of negative reinforcers in Old and Middle French
243
hosting ne . . . mie in the four texts are of this type, while the same holds for between 34% and 50.9% of those marked by ne . . . pas: (33)
(34)
Turpins de Reins, quant se sent abatut, De .IIII. espiez par mi le cors ferut, Isnelement li ber resailit sus ; Rollant reguardet, puis si li est curut, E dist un mot : ) Ne sui mie vencut . . . (Roland, vv. 2083–87) ‘Turpin of Reims, when he finds himself knocked down, his body pierced by four spears, immediately, the brave man stands up again ; he looks around for Roland, then ran to him, and said one word: ‘‘I’m not [mie] defeated . . . ’ . . . si lor avint si merveilleuse aventure qui tuit li huis dou pale´s ou il mengoient et les fenestres clostrent par eles en tel manie`re que nus n’i mist la main ; et neporquant la sale ne fu pas ennuble; (Graal, p. 7) ‘ . . . then there occurred the most marvellous event namely that all the doors and the windows of the palace where they were eating closed themselves without anyone touching them; and yet the hall was not [pas] dark;’
4.3.2. Plain ne. In contrast to the two reinforced forms, the plain ne is used to mark new information in about half the cases in the two early texts in the database, and overwhelmingly so in the two later ones. The discourse status of clauses negated by ne alone is laid out in Table 12.8, and examples given in (35)–(38). (35)
(36)
(37)
) Oez, seignurs, quel pecchet nus encumbret. / Li empereres Carles de France dulce / En cest paı¨s nos est venuz cunfundre. / Jo nen ai ost qui bataille li dunne, (Roland, vv. 15–18) ‘‘‘Hear, Lords, what disaster befalls us. Charles, emperor of sweet France has come to this country to crush us. I don’t have an army that can combat him,’ Car se Franceis te veient entrepiez, / Diront Normant en nom de reprovier : / ) De si fait rei n’avions nos mestier. (Louis, vv. 197–99) ‘For if the French see you trodden down, The Normans will say in reproach: Of such a king we didn’t have need.’ Ceste costume ai je toz jorz tenue et la tendrai tant come je porrai. Mes je avoie si grant joie de Lancelot et de ses cousins qui estoient venu a cort sain et haitie´ qu’il ne me sovenoit de la costume. (Graal, p. 5) ‘This custom have I always observed and I will observe it as long as I can. But I was so happy about Lancelot and his cousins who had arrived at court healthy and safe that I didn’t remember about the custom.’
Table 12.8. Discourse status of clauses negated by the plain preverbal negator Roland (%)
Louis (%)
Graal (%)
Joinville (%)
New information
50
54
68
78
Old information
50
46
32
22
244
Maj-Britt Mosegaard Hansen
(38)
Or avint ainsi que Olivier de Termes, qui bien et vigoureusement s’estoit maintenus outre mer, lessa le roy et demoura en Cypre, lequel nous ne veismes puis d’an et demi apre`s. (Joinville, P16) ‘Then it happened thus that Olivier de Termes, who had kept himself in good and vigorous health overseas, left the King and remained in Cyprus, and we didn’t see him until a year and a half later.’
The 50–50 split between old and new information in the two early texts is what we would expect given that the sole ne was, as discussed in Section 3.3, the unmarked form of negation at that time, and therefore in principle capable of covering both the domain of new information, in contrast to the reinforced negators, and the domain of given information, also covered by the latter. I will not delve further into its use in clauses expressing new information. A more in-depth analysis of the examples featuring plain ne in clauses expressing given information reveals, however, that many of these examples tend to have special semantic features that downplay the discourse salience of the information. Thus, plain ne is often used in irrealis contexts, including but not limited to conditional constructions, where ne can be found in either the antecedent (as in (39), where the underlying proposition represents a paraphrase of an earlier statement) or the consequent (as in (40) which can be analyzed as a paraphrase of, or inference from, the immediately preceding sentence). Example (41), on the other hand, again a paraphrase of a statement made previously, exemplifies a different type of irrealis context, namely a deontic, maxim-like statement, where the negation occurs in a result clause whose content is purely virtual. In all three cases, the bit of previous discourse that the negative clause points back to, and which represented new information at the time it occurred, features the plain negator as well: (39) (40)
(41)
Se cest’ acorde ne volez otrier, / En Sarraguce vus vendrat aseger; (Roland, vv. 475–476) ‘If you will not consent to this agreement, He will come and besiege you in Saragoza;’ Envers le povre te deis umeliier ; / Se il se claime, ne te deit enoier. (Louis, vv. 182–183) ‘Towards the poor man you must be humble; if he complains, you must not be annoyed.’ Car, ce dit li saiges, on se doit assemer en robes et en armes en tel maniere que li preudome de cest siecle ne dient que on en face trop, (Joinville, P38) ‘For, so says the wise man, one must equip oneself with clothes and arms in such a way that the serious men of this world shall not say that one is overdoing it,’
Other examples feature ne negating a modal verb, in contexts which resemble the irrealis contexts exemplified above. Thus, in (42), which represents a highly salient inference from the preceding text, Ganelon is in fact where he doesn’t want to be: (42)
Li empereres li tent sun guant, le destre ; / Mais li quens Guenes iloec ne volsist estre. (Roland, vv. 331–332)
The grammaticalization of negative reinforcers in Old and Middle French
245
‘The emperor holds out his glove, the right one, to him; but Count Ganelon would rather not have been there.’ We also encounter the plain ne in non-declarative clauses like the presuppositional (43), and in non-referential contexts like (44), which expresses an inference from the immediately previous text: (43) (44)
) Pourquoy ne prenez-vous ce que nos gens vous offrent? * (Joinville, P58) ‘‘‘Why don’t you accept what our people are offering you?’’’ Si i acorrent li un et li autre en tel maniere qu’il ne remest chevalier en tot le pale´s qui la ne venist. (Graal, p. 11) ‘Thus they come running from various places in such a way that there was no knight left in the whole palace who didn’t come there.’
The question in (43), however, focuses not on the fact that the addressee does not accept what he is offered, but rather on the reasons for his refusal to do so. The negated proposition is thus communicatively backgrounded here. As for (44), we find a sequence of two negatives (‘‘there is no X that does not F’’), which together express a positive statement (‘‘all X F’’). This type of structure is not infrequent in my data, which also feature a few other cases, such as the paraphrastic (45), where a negative clause occurs in an approximative construction which, as a whole, focuses on the fact that the opposite, positive, state-of-affairs came very close to being realized: (45)
Quant c- o veit Guenes que ore s’en rit Rollant, / Dunc ad tel doel pur poi d’ire ne fent ; / A ben petit que il ne pert le sens, (Roland, vv. 303–305) ‘When Ganelon sees that Roland is laughing at it now, he hurts so much that he almost bursts with anger; He almost loses his mind,’ (lit.: it’s due to a very small thing that he doesn’t lose his mind)
Several of these features may at times be combined in one and the same example (e.g., (40) above, which features a conditional construction and a deontic modal verb, or (46), in which we find a conditional construction with a non-declarative main clause), and all of them can be found in ne-marked clauses expressing new information, as well. Indeed, some of the contexts mentioned (with modal verbs, in conditionals introduced by the conjunction si, and in a subordinate clause following a negated main clause) are precisely the ones in which one may, in formal registers, encounter the simple preverbal ne even in Modern French, cf. (47). (46) (47)
N’en parlez mais, se jo nel vos cumant! (Roland, v. 273) ‘Don’t speak any more, if I don’t command you to do so!’ Il n’y a rien a` voir dans ce quartier, si ce n’est des immeubles gris des anne´es soixantedix. ‘There’s nothing to see in this neighborhood, apart from (lit.: if it is not) grey apartment buildings from the seventies.’
246
Maj-Britt Mosegaard Hansen
4.3.3. ‘‘Janus-faced’’ examples of ne . . . mie/pas. I hope to have convincingly established in the above that there were, indeed, discourse–functional constraints of the hypothesized nature on the use of reinforced negation in Old and Middle French. It remains now to propose an explanation of why these constraints were gradually lifted, allowing ne . . . pas to grammaticalize as the new standard negation, fully generalized at the beginning of the Classical French period in the 17th century. In a quantitative study of the use of competing negative forms in a 14th text, Le Roman de Be´rinus, Offord (1976: 333) observes that negative reinforcers are very frequently used in adversative contexts marked by the conjunctions mes (‘but’) and (the more strongly contrastive) ainz. Indeed, my data show the same kind of pattern, cf. (48)–(49): (48)
(49)
)Biau Sire, fet Gauvains, donc me poez vos bien dire, s’il vos plest, en quoi sui tiex come vos me metez sus. *-) Je nel vos dirai mie, fet cie, mes vos troveroiz par tenz qui le vos dira. * (Graal, p. 52) ‘‘‘Good Sir, says Gawain, then you can surely tell me, if you please, in what way I am that which you accuse me of.’’ – ‘‘I’ll not [mie] tell you, he says, but soon you’ll find one who’ll tell you.’’’ Quant li lyons se voit delivrez dou serpent par l’aide dou chevalier, si ne fait pas semblant que il ait talent de combater a lui, ainz vient devant lui et besse la teste . . . (Graal, p. 97) ‘When the lion finds itself freed from the serpent by the knight’s help, it does not [pas] seem to want to fight him, but comes before him and bows its head . . . ’
Offord (1976: 333) suggests that this is so because the contrastive meaning of the conjunctions in question emphasizes the negative aspect of the first conjunct. However, a closer look at the examples shows that the use of the reinforced negators can, in fact, be explained as governed by the discourse constraints detailed in Section 3. Thus, (48) is a denial of Gawain’s immediately preceding request, while (49) denies an expectation based on an earlier statement to the effect that the knight knows the wild beasts will kill him if they get a chance. Nevertheless, I will argue that this type of example – and others like it, but which do not involve adversative conjunctions – is likely to have been very important in the grammaticalization of ne . . . pas, for the contexts of the negations in (48)–(49) have the peculiar property of being Janus-faced. That is to say that, while the negated clauses are backwards oriented in the sense of being discourse-old, they are at the same time forwards oriented in the sense of expressing a contrast with the contents of the immediately following clause. Largely similar examples can be found where ne . . . mie/pas occur in explicitly causal contexts such as (50): (50)
Et quant len li volt demander qui il estoit, il n’en tint onques plet a ax, ainz respondi tot pleinement qu’il ne lor diroit ore pas, car il le savroient bien a tens se il l’osoient demander. (Graal, p. 8)
The grammaticalization of negative reinforcers in Old and Middle French
247
‘And when they tried to ask him who he was, he never exchanged words with them, but replied quite plainly that he wouldn’t [pas] tell them now, for they’d soon find out if they dared to ask.’ Finally, there are a handful of examples in my database that are Janus-faced in the sense that an inference is being either expressed or denied, but while that inference is in each case partially warranted by the preceding text, it nevertheless appears to be strengthened by further support in the immediately following text. Thus, with respect to (51), while the fact that the sun does not shine in a given region certainly does suggest that the land may not be very fertile, the inference that corn does not grow there is a mere bridging inference, and although accessible based on the information given in the preceding text, it cannot be said to be very salient in the discourse prior to its utterance. It is greatly strengthened, however, when further information is added about the dryness of the climate and the stoniness of the land. (51)
Icele tere, c- o dit, dun il esteit, / Soleill n’i luist ne blet n’i poet pas creistre, / Pluie n’i chet, rusee n’i adeiset, / Piere n’i ad que tute ne seit neire: (Roland, vv. 979-82) ‘In that country it is said, where he was from, the sun doesn’t shine, and corn can’t [pas] grow, the rain doesn’t fall, the dew doesn’t touch it, there is no stone that isn’t all black:’
It is highly plausible that Janus-faced contexts constitute the crucial bridging contexts for the reanalysis of ne . . . mie/pas, because they leave the addressee free to interpret the reinforced negator either as fulfilling its original function of displaying a backwards orientation toward a contextually given proposition, or as fulfilling a new function of displaying forwards orientation toward a proposition that is yet to be uttered. If this is correct, examples like (52), of which there are a couple in La Queste del Saint Graal and in Joinville, can be seen as what Heine (2002: 85) calls ‘‘switch’’ contexts, i.e., contexts where only the innovative meaning or function – in casu the forwards orientation – makes for a plausible interpretation: (52)
Quant Perceval se fu endormiz, si li avint une aventure merveilleuse : car il li fu avis en son dormant que devant lui venoient deus dames dont l’une ert vielle et ancienne et l’autre n’ert mie de mout grant aage, mes bele estoit. Les deus dames ne venoient pas a pie´, ainz estoient montees sus deus molt diverses bestes: . . . (Graal, p. 97) ‘When Percival had fallen asleep, a miraculous event occurred: for it seemed to him as he was sleeping that two ladies came before him, one of whom was very old and the other wasn’t of very old age, but was beautiful. The two ladies didn’t [pas] come on foot, but were riding two very strange beasts: . . . ’
Here, there appears to be no very good reason for the reader to assume that the two ladies would be using any particular means of locomotion, as opposed to others, so the negated clause appears oriented only toward expressing information that contrasts with that in the following clause. In other words, the connection to prior discourse that I have argued is standardly signaled
248
Maj-Britt Mosegaard Hansen
by the use of reinforced negation in older stages of the French language, has been severed here. Once such switch contexts start to occur, bi-partite negation, which was originally a conceptually marked construction, is ready to become conceptually unmarked, i.e., essentially indistinguishable from a generalized negator, which can occur in both backwards and forwards oriented contexts.
5. CONCLUSION This paper has presented evidence that key stages in the diachronic evolution of clausal negation in French should be understood as governed by discourse–functional constraints on the flow of information. Specifically, I have argued, concerning the synchronic properties of Old and Middle French negation, that clauses negated by ne . . . mie/pas were constrained to be discourseold, as defined by Birner (2006), and that, while the proposition expressed by such clauses need not be believed, it should be such that the speaker could assume that it was either already activated in the short-term memory of the hearer or accessible to activation based on other propositions thus activated. This analysis presents the advantage of being compatible with what is known about the uses of different forms of negation in a number of contemporary Romance vernaculars where variation is still maintained between simple preverbal negators and reinforced expressions which in several cases are etymologically identical to the French forms. I have suggested, further, a diachronic scenario capable of explaining the subsequent unmarking of reinforced negation in French, my proposal being that so-called Janus-faced contexts, i.e., contexts that are at one and the same time backwards and forwards oriented in terms of the flow of information in discourse, constituted the key bridging contexts that allowed for the reanalysis of the reinforced negators. The advantages of the proposed scenario are that it not only relies on precisely those discourse-functional constraints that were argued to govern the use of negative reinforcers at the stage where they were still conceptually and textually marked, but it is also more precise than existing pragmatically based explanations.
ACKNOWLEDGMENT I wish to thank Adrian Armstrong for helpful comments on a draft version of this paper. Needless to say, he is not responsible for any errors contained in the presentation.
REFERENCES Ashby, W. J. (1981). ‘‘The loss of the negative particle ne in French: A syntactic change in progress’’. Language 57 (3): 674–687.
The grammaticalization of negative reinforcers in Old and Middle French
249
Ashby, W. J. (2001). ‘‘Un nouveau regard sur la chute du ne en franc- ais parle´ tourangeau: s’agit-il d’un changement en cours ?’’. Journal of French Language Studies 11: 1–22. Bernini, G. and P. Ramat (1996). Negative sentences in the languages of Europe. A typological approach. Berlin: Mouton de Gruyter. Birner, B. J. (2006). Semantic and pragmatic contributions to information status, in M.-B. Hansen Mosegaard and K. Turner (eds.), Explorations in the semantics/pragmatics interface, Special issue of Acta linguistica Hafniensia 38: 14–32. Bybee, J., R. Perkins and W. Pagliuca (1994). The evolution of grammar. Tense, aspect, and modality in the languages of the world. Chicago: University of Chicago Press. Coveney, A. (1996). Variability in spoken French. A sociolinguistic study of interrogation and negation. Exeter: Elm Bank Publications. ¨ . (1979). ‘‘Typology of sentence negation’’. Linguistics 17: 79–106. Dahl, O Detges, U. and R. Waltereit (2002). ‘‘Grammaticalization vs reanalysis: A semantic-pragmatic account of functional change in grammar’’. Zeitschrift fu¨r Sprachwissenchaft 21 (2): 151–195. Dryer, M. S. (1996). ‘‘Focus, pragmatic presupposition, and activated propositions’’. Journal of Pragmatics 26: 475–523. Eckardt, R. (2003). The structure of change. Meaning change under reanalysis, Habilitation-thesis. Berlin: Humboldt Universita¨t. (Revised version published by Oxford University Press, 2006). Espinal, M. T. (1993). ‘‘The interpretation of no-pas in Catalan’’. Journal of Pragmatics 19: 353–369. Evans, N. and D. Wilkins (2000). ‘‘In the mind’s ear: The semantic extensions of perception verbs in Australian languages’’. Language 76 (3): 546–592. Fonseca-Greber, B. B. (2007). ‘‘The emergence of emphatic ne in conversational Swiss French’’. Journal of French Language Studies 17 (3): 249–275. Foulet, L. (1965). Petite syntaxe de l’ancien franc- ais. Paris: Honore´ Champion. Givo´n, T. (1976). ‘‘Topic, pronoun, and grammatical agreement’’, in C. N. Li (ed.), Subject and topic. New York: Academic Press, 149–188. Givo´n, T. (1990). Syntax. A functional-typological introduction. Vol. II. Amsterdam: John Benjamins. Guiraud, P. (1964). ‘‘L’opposition actuel-virtuel. Remarques sur l’adverbe de ne´gation dans Aucassin et Nicolette’’, in J. Renson, J. Horrent, L. Remacle and M. Tyssens (eds.), Me´langes de linguistique romane et philologie me´die´vale offerts a` M. Maurice Delbouille. Vol. I. Gembloux: J. Duculot, 295–306. Hansen, A. B. and I. Malderez (2004). ‘‘Le ne de negation en re´gion parisienne : une e´tude en temps re´el’’. Langage et Socie´te´ 107: 5–30. Hansen, M.-B. Mosegaard (2008). Particles at the semantics/pragmatics interface: Synchronic and diachronic issues. A study with special reference to the French phasal adverbs (Current research in the semanticspragmatics interface, Vol. 19 ). Oxford/Bingley: Elsevier/Emerald. Hansen, M.-B. Mosegaard and J. Visconti (2009). ‘‘On the diachrony of ‘‘reinforced’’ negation in French and Italian’’, in C. Rossari, C. Ricci and A. Spiridon (eds.), Grammaticalization and pragmatics: Facts, approaches, theoretical issues (Studies in Pragmatics, Vol. 5). Bingley: Emerald, 137–171. Harris, M. (1978). The evolution of French syntax. London: Longman. Heine, B. (2002). ‘‘On the role of context in grammaticalization’’, in I. Wischer and G. Diewald (eds.), New reflections on grammaticalization. Amsterdam: John Benjamins, 83–101. Jespersen, O. (1917). Negation in English and other languages (Det Kgl. Danske Videnskabernes Selskab. Historisk-Filologiske Meddelelser I, Vol. 5). Copenhagen: Høst & Søn. Lehmann, W. P. (1973). ‘‘A structural principle of language and its implications’’. Language 49 (1): 47–66. Marchello-Nizia, C. (1999). Le franc- ais en diachronie : douze sie`cles d’e´volution. Paris: Ophrys.
250
Maj-Britt Mosegaard Hansen
Martin, R. (1972). ‘‘La ne´gation de ‘‘virtualite´’’ du moyen franc- ais’’. Romania 93: 20–49. Martineau, F. and R. Mougeon (2003). ‘‘A sociolinguistic study of the origins of ne deletion in European and Quebec French’’. Language 79 (1): 118–152. Mo¨hren, F. (1980). Le renforcement affectif de la ne´gation par l’expression d’une valeur minimale en ancien franc- ais. Tu¨bingen: Niemeyer. Offord, M. (1976). ‘‘Negation in ‘‘Be´rinus’’: A contribution to the study of negation in fourteenth century French’’. Zeitschrift fu¨r Romanische Philologie 92: 313–385. Perle, F. (1878). ‘‘Die negation im altfranzo¨sischen’’. Zeitschrift fu¨r romanische Philologie II: 1–24. Price, G. (1962). ‘‘The negative particles pas, mie and point in French’’. Archivum linguisticum 14 (1): 14–34. Price, G. (1990). ‘‘The origins and syntax of ne . . . goutte’’, in J. N. Green and W. Ayres-Bennett (eds.), Variation and change in French. Essays presented to Rebecca Posner on the occasion of her sixtieth birthday. Routledge, London, 201–209. Price, G. (1993). ‘‘Pas (point) without ne in interrogative clauses’’. Journal of French Language Studies 3: 191–195. Prince, E. F. (1992). ‘‘The ZPG letter: Subjects, definiteness, and information-status’’, in S. A. Thompson and W. Mann (eds.), Discourse description: Diverse analyses of a fund-raising text. Amsterdam: John Benjamins, 295–325. Manuscript downloaded from http://www.ling.upenn.edu/Bellen/home.html Schwegler, A. (1988). ‘‘Word order changes in predicate negation strategies in Romance languages’’. Diachronica 5: 21–58. Schwenter, S. A. (2006). ‘‘Fine-tuning Jespersen’s cycle’’, in B. J. Birner and G. Ward (eds.), Drawing the boundaries of meaning. Neo-Gricean studies in pragmatics and semantics in honor of Laurence R. Horn. Amsterdam: John Benjamins, 327–344. Stalnaker. R. C. (1991[1974]), ‘‘Pragmatic presuppositions’’, in S. Davis (ed.), Pragmatics. A reader. Oxford: Oxford University Press, 471–481. Sten, H. (1938). Nægtelserne i fransk. En historisk-syntaktisk fremstilling. Copenhagen: Nyt Nordisk Forlag. Togeby, K. (1974). Pre´cis historique de grammaire franc- aise. Copenhagen: Akademisk Forlag. Van der Auwera, J. and A. Neuckermans (2004). ‘‘Jespersen’s cycle and the interaction of predicate and quantifier negation in Flemish’’, in B. Kortmann (ed.), Dialectology meets typology. Berlin: Mouton de Gruyter, 453–478. Vennemann, T. (1974). ‘‘Topics, subjects, and word order: From SXV to SVX via TVX’’, in J. M. Anderson and C. Jones (eds.), Historical linguistics I. Syntax, morphology, internal and comparative reconstruction. Amsterdam: North-Holland, 339–376. Winters, M. E. (1987). ‘‘Innovations in French negation. A cognitive grammar account’’. Diachronica IV (1–2): 27–52. Yvon, H. (1948). ‘‘Pas et point dans les propositions ne´gatives’’. Le franc- ais moderne XVI: 19–35. Zanuttini, R. (1997). Negation and clausal structure. A comparative study of the Romance languages. Oxford: Oxford University Press.
DATA SOURCES BFM ¼ Base de franc- ais me´die´val, searchable electronic data basehttp://weblex.ens-lh.fr/wlx/cgi/weblexf ?corpus ¼ bfm La chanson de Roland. (Ge´rard Moignet, ed.). (1985[1080]), Bordas, Paris.
The grammaticalization of negative reinforcers in Old and Middle French
251
Le couronnement de Louis. (Ernest Langlois, ed.). (1961[ca. 1150]), Honore´ Champion, Paris. Joinville, Jean Sire de. (1874[1298–1309]), Histoire de Saint Louis. Credo et Lettre a` Louis X (Natalis de Wally, ed.), Firmin Didot Fre`res, Fils et Co., Paris. LLT ¼ Library of Latin Texts, searchable electronic data basehttp://clt.brepolis.net/clt/start.asp?sOwner ¼ menu La Queste del Saint Graal (Albert Pauphilet, ed.). (1967[ca. 1220]), Champion, Paris. ‘‘Voyage en Egypte’’, conversation recorded and transcribed by Michel Bichard, Universite´ de la Sorbonne Nouvelle, Paris 3, 1991–1992.
Current Trends in Diachronic Semantics and Pragmatics Edited by Maj-Britt Mosegaard Hansen and Jacqueline Visconti r 2009 by Emerald Group Publishing Limited. All rights reserved.
13 A ROOTS JOURNEY
OF A
FRENCH PREPOSITION
Silvia Adler and Maria Asnes
1. INTRODUCTION This study traces the origins of the French preposition jusqu’a` ‘until’, ‘up to’, ‘to the point of’, ‘to’. Our roots journey opens with a synchronic analysis of the various usages of this preposition before delving into the historical investigation. We intend to show that even though the spatiotemporal readings chronologically slightly precede other possible usages, all the values of jusqu’a` already coexist at the early stages of French, as well as in Latin. This fact may confirm the hypothesis of the monosemic nature of this preposition. In other words, the spatio-temporal meaning of jusqu’a` exists alongside scalar and quantificational meanings, and thus cannot be considered as the core meaning. In fact, all the possible readings of PPs headed by jusqu’a` share one semantic primitive which has to do with the notions of path and of a culminating point (representing the limit of the path). What actually distinguishes the different readings of jusqu’a` is the nature of the limit provided by the context. In Section 2, all possible values of jusqu’a` will be analyzed from a synchronic perspective. Section 3 will portray the diachronic trajectory, based on empirical data from electronic corpora, mainly from the Base de Franc- ais Me´dieval. This will support the idea of the importance of context for the interpretation process. Finally, Section 4 will motivate the semantic diversity of contexts in which this preposition appears.
254
Silvia Adler and Maria Asnes
2. SYNCHRONIC MEANINGS Synchronically, the French preposition jusqu’a` has four major usages: spatial, temporal, scalar, and quantificational. 2.1. Spatio-temporal reading According to the Tre´sor de la Langue Franc- aise (TLF), jusqu’a` expresses the idea of a spatiotemporal limit (for space or duration): 2.1.1. Spatial reading. (1) a. Pousser la voiture jusqu’au garage ‘To push the car to the garage’ b. Arriver jusqu’au garage ‘To arrive to the garage’ c. *Se trouver jusqu’au garage ‘To be located to the garage’ In general, spatial PPs introduced by jusqu’a` modify locative predicates. Nonetheless, there is a restriction on the aspectual type1 of the modified predicate. Jusqu’a`, which contains both directionality and a terminal point in its denotation, is compatible with activities (1a) and achievements (1b): activities imply duration and permit directionality, while achievements imply reaching a terminal point. It is thus incompatible with states (1c) which lack both directionality and a terminal point.2 State jusqu’a`
Activity þ
Accomplishment Not applicable (NA)
Achievement 3
þ
1
See Vendler (1967). This claim is true for static states (in the sense of Bach, 1986). However, dynamic states, implying directionality, can be compatible with jusqu’a` (e.g., s’e´tendre jusqu’a`). 3 Accomplishments are mostly the result of adding a PP headed by jusqu’a` to activities, as the test involving en (in)/pendant (for) applied to the activity predicate courir (to run) may confirm: 2
courir jusqu’au garage courir pendant/*en 2 heures BUT: courir jusqu’au garage en 2 heures The combination of the activity predicate courir and a PP headed by jusqu’a` results in an accomplishment.
A roots journey of a French preposition
255
2.1.2. Temporal reading. (2) a. Durer jusqu’a` minuit ‘To last till midnight’ b. Travailler jusqu’a` minuit ‘To work till midnight’ c. Construire une maison jusqu’au mois de mai ‘To build a house till May’ d. Terminer le projet jusqu’a` 20:00 ‘To finish the project till 8 PM’ We can see that temporal PPs headed by jusqu’a` are compatible with Vendlerian states (2a) and activities (2b) due to their durational nature, and with accomplishments (2c) and achievements (2d) given the presence of an endpoint in their denotation.
jusqu’a`
State
Activity
Accomplishment
Achievement
þ
þ
þ
þ
2.2. Scalar reading The TLF also points out that jusqu’a` can denote an extreme degree or a limit which cannot be surpassed. PPs headed by jusqu’a` can modify: a. psychological predicates (3) aimer/eˆtre aime´ jusqu’a` l’adoration ‘to love/be loved to the point of worshipping’ b. activities (4) crier jusqu’a` l’e´puisement ‘to shout to the point of exhaustion’ c. degree achievements (5) maigrir jusqu’a` l’autodestruction ‘to lose weight to the point of auto-destruction’ d. individual- and stage-level predicates (6) eˆtre froid jusqu’a` l’impolitesse ‘to be cold to the point of impoliteness’ e. open-scale predicates (7) ravi jusqu’a` l’extase ‘charmed to the point of ecstasy’
256
Silvia Adler and Maria Asnes
f. closed-scale predicates (8) remplir jusqu’a` e´clatement ‘to fill to the point of explosion’ g. predicates of motion in an abstract reading (9) il est alle´ jusqu’a` vendre la maison de ses parents ‘he has gone to the point of selling the house of his parents’ (for example, said about someone who gambles) PPs headed by jusqu’a` function as predicate intensifiers and, as such, imply a boundary of an intensional nature, as opposed to spatio-temporal readings where the boundary is basically extensional.4 The DP in the scope of jusqu’a` denotes a culminating point or a final boundary reached by the predicate’s referent. The intensification is thus bounded (cf. Adler and Asnes, 2004, 2010b), although, pragmatically speaking, one uses utterances such as these to denote, first of all, the high degree of the modified predicate and even the fact that this intensity surpasses a pre-established standard. 2.3. Quantificational reading According to the TLF, in examples such as (10a) jusqu’a` is a synonym of meˆme (even) and y compris (included/including) and it denotes a limit for the inclusion in a set:5 (10)
a. Des femmes et jusqu’a` des enfants travaillaient aux barricades ‘Women and up to (even) children were working at the barricades’ (Zola, De´baˆcle, 1892, p. 594).
In this additive reading, jusqu’a` precedes a noun phrase, and together they function as a noun phrase and not as a prepositional phrase. This explains why the PP headed by jusqu’a` can occupy subject and direct object positions. Moreover, the preposition may be omitted here without affecting the grammaticality of the sentence, as example (10b) suggests: (10)
4
b. Des femmes et des enfants travaillaient aux barricades ‘Women and children were working at the barricades’
We refer here to the intension/extension dichotomy, as defined in logic and semantics. In our case, an intensional boundary refers to the limit imposed to the property denoted by the predicate, whereas an extensional boundary denotes a limit imposed on the predicate’s referent. 5 Contrary to (11), jusqu’a` in (10a) does not precede a quantified noun phrase. We include (10a) in this section because of the common syntactic properties it shares with (11), i.e., the fact that jusqu’a` þ DP functions as a DP and not a PP. According to TLF, this reading is already present in the 16th century.
A roots journey of a French preposition
257
However, the TLF does not mention contexts in which jusqu’a` precedes quantified noun phrases, although these examples are of a very frequent use in Present Day French (11): (11)
a. Remplir jusqu’a` 5 demandes ‘To fill in up to 5 applications’ b. . . . 5 demandes tout au plus ‘ . . . at most 5 applications’
Surprisingly, the current linguistic literature tends not to associate this preposition with the quantificational domain. In this usage, jusqu’a` functions as a modifier of quantifiers similarly to au plus (at most) and au moins (at least). Thus, in (11a) jusqu’a` can be paraphrased by tout au plus (at most) (11b). In other words, jusqu’a`, similarly to tout au plus, triggers a superlative reading by marking the upper boundary of a certain quantity (r). The paraphrases, for instance n au plus (where n represents a numeral), are usually categorized as modified cardinals.6 However, this is not the case of jusqu’a`, whose syntactic categorization remains problematic. For instance, the mobility criterion (cf. Dobrovie-Sorin and Beyssade, 2004) excludes jusqu’a` from the category of adverbs (see examples (12)–(15)): (12) (13) (14) (15)
Je pourrais vous vendre (tout) au plus 50 caisses ‘I could sell you at most 50 boxes’ Je pourrais vous vendre jusqu’a` 50 caisses ‘I could sell you up to 50 boxes’ Au plus, je pourrais vous vendre 50 caisses ‘At most, I could sell you 50 boxes’ *Jusqu’a`, je pourrais vous vendre 50 caisses *‘Up to, I could sell you 50 boxes’
Contrary to the abovementioned adverbs, jusqu’a` cannot modify finite verbs, adjectives, or sentences ((16)–(19)): (16) (17) (18) (19)
6
Il s’est au moins excuse´ ‘He has at least apologized’ Au moins il s’est excuse´ ‘At least, he has apologized’ Il est au moins poli ‘He is at least polite’ Au moins il est poli ‘At least he is polite’
cf. Krifka (1999).
258
Silvia Adler and Maria Asnes
The paraphrases involving jusqu’a` are not possible. If jusqu’a` is not an adverb, what is it then? Does it behave as a real preposition in this particular reading? This seems not to be the case either. For example, jusqu’a` can be omitted without affecting the grammaticality of the sequence. ((11c) is a modification of (11a)): (11)
c. Remplir 5 demandes ‘To fill in 5 applications’
This is not possible for the other readings in which jusqu’a` is the head of real prepositional phrases (as shown by (20) with a temporal reading and (21) with a scalar reading):7 (20)
(21)
Travailler jusqu’a` 3 heures du matin *Travailler 3 heures du matin *‘To work 3 o’clock in the morning’ Maigrir jusqu’a` l’autodestruction *Maigrir l’autodestruction *‘To lose weight auto-destruction’7
Thus, what seems to have the same surface structure corresponds to two different analyses: (22) (23)
Chanter [PP jusqu’a` [DP 3 heures du matin]] ‘To sing [PP until [DP 3 o’clock in the morning]]’ Chanter [DP [QP jusqu’a` 2] notes diffe´rentes] en meˆme temps ‘To sing [DP [QP up to 2] different notes] at the same time’
In (22), the preposition jusqu’a` has in its scope the complement 3 heures du matin, and its omission is not possible: the preposition is there to establish a relation between the predicate chanter and the noun phrase 3 heures du matin. On the other hand, jusqu’a` in (23) has no relation whatsoever with chanter, but has in its scope only the numeral 2. What happens with the quantificational readings involving jusqu’a` is that an item belonging to a lexical category of prepositions loses the essence of this category, which is serving as a relational word, and acquires a new function. Jusqu’a` forms together with the numeral a quantificational phrase (QP), in which it functions as a modifier of a quantifier.8
7 For other tests proving that in this reading jusqu’a` is neither a typical preposition nor an adverb, see Adler and Asnes (2010a). 8 Cf. Melis (2001) regarding the different categorization strategies of prepositions with atypical behavior. Cf. also Ilinski (2003).
A roots journey of a French preposition
259
3. DIACHRONIC EVOLUTION We argue that all of the currently coexisting readings are already present at the early stages of French as well as in Latin. Thus, we do not have here a typical case of evolution from a concrete, spatial core to all other abstract meanings. Instead, we will put forth the hypothesis that we are dealing here with a case of one semantic primitive, common to all the readings, which is related to the notion of a path and its limit. Space, time, degree, and quantification seem to be particular instances of this basic sense core. Let us point out that in Old French, jusque exists along with a synonymous preposition that eventually disappeared: tresque. This preposition derived from the preposition tres and the conjunction que. Similarly to jusqu’a`, tresque marks a spatio-temporal limit (see example (24) in which it is told that Charles the king conquered the highland up to the sea): (24)
Carles li reis, nostre emper[er]e magnes Set anz tuz pleins ad estet en Espaigne: Tresqu’en la mer cunquist la tere altaigne. ‘Charles the king, our great emperor, Has stayed seven whole years in Spain, And has conquered the haughty country as far as the sea’ (Chanson de Roland 1090, verse 1, translation by W. S. Merwin)
This section first traces the etymology of jusqu’a` in order to reveal its primary meaning. Then, through a series of examples taken from the BFM database (Base du Franc- ais Me´die´val), all the synchronic usages in question are traced back to Old French. 3.1. Etymology A survey of different etymological dictionaries9 reveals that jusque (josque) derives from the popular Latin de usque, de usquam, which, in its turn, originates from classical Latin usque. In classical Latin usque means ‘all the way, up to, even (to)’ and usquam means ‘anywhere’. According to the Dictionnaire Etymologique de la langue franc- aise,10 Old French jusque denotes reinforcement just like the Latin usque. In Modern French, jusqu’a` is a compound preposition formed from two simple prepositions: jusque and a` and it is derived from Latin (ode usque ad).
9
Dauzat (1974); Bloch and von Wartburg (1975). Dauzat (1974).
10
260
Silvia Adler and Maria Asnes
Following Lehmann (1995 [1982]: 86), we suggest that in the case of usque ad we have an instance of reinforcement of the weakened Latin preposition ad by the emphatic adverbial particle usque. The first question that arises concerns the status of usque in Latin. From the syntactic point of view, usque is analyzed as an adverb which can also serve as a prepositional specifier. The possibility of omission of usque confirms its syntactic redundancy.11 In addition to its syntactic redundancy, its position in the sentence is free: usque can precede the preposition it specifies (25) or be separated from it (26):12
11
If cross-linguistic data can consolidate the idea of redundancy, let us draw attention to some synchronic examples from Hebrew, Italian, Russian, and German. All these examples show that simple monosyllabic prepositions can be intensified by another particle or preposition: 1. In Hebrew the complex French preposition jusqu’a` in all its usages translates into the simple preposition ‘ad’. However, in the scalar reading there is a possibility to intensify it by another preposition – kdei – which usually conveys the meaning of goal. In conjunction with ‘ad’ it seems to lose its original meaning and turns into a simple emphatic particle. It is not surprising, though, that a preposition denoting a goal is selected to intensify ‘ad’ since both are telic: Leehov ad dma’ot/ leehov ad kdei dmao’ot Love to tears/love up to tears 2. In Italian, the complex preposition fino a can be used for spatio-temporal, scalar and quantificational readings. However, in the scalar reading fino can be omitted: Commuovere alle lacrime/Emozionare fino alle lacrime Move to tears/Touch up to tears Similarly to Hebrew kdei, fino conveys the idea of goal, cf. Visconti (2005a, b). 3. In Russian, jusqu’a` translates into a simple preposition do in all its usages. Yet, there is a general possibility to intensify do by an emphatic particle azˇ. For instance: Quantificational: Priniat’ do 1500 cˇelovek/Priniat’ azˇ do 1500 cˇelovek To admit up to 1500 people/To admit up to 1500 people Scalar:
Liubit’ do slioz/Liubit’ azˇ do slioz Love to tears/love up to tears
4. German bis (to) and zu (to) convey roughly the same meaning, while their combination (bis zu) leads to intensification: bis Tra ¨ nen lachen/ zu Tra¨nen ru¨hren/bis zu Tra¨nen lachen Love to tears/Touch to tears/Love up to tears
12
The examples are borrowed from Visconti (2004) following Vincent (1999: 1136).
A roots journey of a French preposition
(25) (26)
261
a coelo usque ad centrum ‘From the sky [heavens] all the way to the center’ ab Attica ad Thessaliam usque ‘From Attica right to Thessaly’
Vincent (1999) claims that from a free adverb, usque evolves into a fixed pre-prepositional, thus functioning as a specifier of PP. Later, the PP in the scope of usque becomes its complement (27): (27)
[P [P DP]]-[P-P [DP]] [usque [ad DP]]-[usque ad [DP]]
The latest stage of this process of reanalysis gives rise to a complex preposition. What is of a particular interest for us is the fact that already in Latin, according to Vincent (1999), this adverb functions as a degree modifier just as the particle right in English (cf. (26)). If so, is usque really spatio-temporal in its origin? The Canadian official motto since 1921 will help us clarify this point: (28)
(et dominabitur) a mari usque ad mare (Old Testament, Psalm 72, St. Jerome’s Latin version of Holy Writ, the Vulgate) ‘and he shall have dominion from one sea all the way through to another sea’
According to Casselman (1995), the King James translation of this verse is: ‘and he shall have dominion also from sea to sea’, where usque is mistranslated as ‘also’. This mistranslation results from the fact that usque, preposed to the preposition ad, emphasizes the meaning of this preposition. The more appropriate translation would be, according to Casselman: ‘from sea yea onto sea’ or ‘from one sea all the way through to another sea’. In this motto, ad alone conveys the sense of the Modern French jusqu’a`, whereas usque behaves as an emphatic particle. This can be opposed to the evolution of perfino and addirittura in Italian (Visconti, 2004, 2005a, b) where a spatio-temporal item evolves into an emphatic particle (or a discourse marker). In our case, the starting point of the semantic evolution seems to be an emphatic degree modifier – in Latin – which becomes – in Modern French – a complex preposition, that is to say a nonomissible item (except the quantificational reading discussed below). The Latin examples show clearly the spatial nature of usque (it is combined with locative DPs). Nonetheless, the spatial meaning already contains an emphatic, subjective dimension. The origin of jusqu’a` already discloses its potential to assume scalar readings: usque, meaning ‘as far as, all the way, continually, straight on, up to, until’, promotes the idea of a continuum, a notion closely related to scalarity. Moreover, this allows us to put forth a hypothesis according to which jusqu’a` is not polysemic, but rather all its readings are realizations of one basic semantic core.
262
Silvia Adler and Maria Asnes
3.2. Semantic evolution13 3.2.1. Spatial. The first occurrence of jusqu’a` can be traced back to the second half of the 10th century: (29)
. . . Et mult corps sanz en sun exut et inter omnes sunt vedut; qui in cortine pend, jusche la terra per mei fend. (Anonymous, La Passion de Clermont, end of 10th century, verse 328) ‘ . . . And many bleeding corpses were taken out of there and seen among people; the one who hangs in bed curtains; breaks down to earth in front of me.’
In this example, jusche precedes the locative DP la terra and thus has a purely spatial reading. Not only does the entire PP have a spatial reading but it also conveys the directionality of the movement (here: from top to bottom).14 3.2.2. Temporal. In the same period, the preposition jusque can be found alone (30) or combined with other prepositions, mostly a` (31): (30)
(31)
Tanz dis se repose et se taise jusque il voie bien son aise (Anonymous, Roman de The`bes (vol. 1) ca. 1150, verse 3890) ‘Meanwhile he rests and is silent till he feels at ease.’ irez ; se il ont roi , a lui parlez , de moie part le saluez et si li dites de ma part que des cors rendre ne se tart , et s’ il nes rent isnelement , jusqu’ a tierz jour venra ma gent , je l’ aserrai en sa cite´. (Anonymous, Roman de The`bes (vol. 2) ca. 1150, verse 10368) ‘go, if they have a king, talk to him, greet him on my behalf and tell him not to be slow to return the bodies, and if he does not return them quickly, until the third day my people will come, I will besiege him in his city’.
3.2.3. Temporal þ scalar. Already in the 12th century, we find temporal readings which contain a scalar dimension due to the scalar nature of the modified predicate: (32)
13
14
Tout autresi fera la moie : jusqu’ a tierz jor ravra sa joie. (Anonymous, Roman de Renart (branch 1), starting from 1180, verse 2066) ‘ . . . till the third day his joy will delight (ravish).’
The history of French is divided into the following periods: 842–1130 – Old French (9th century – 1st half of the 12th century) mid-12th century – beginning 14th centuries – Classical French 14th–16th centuries – Middle French We do not quote examples from Classical French since the same usages persist during this period. In general, our purpose is to discover the origins of jusqu’a` rather than to trace its entire evolution through the centuries. Here jusche is not accompanied by the preposition a`.
A roots journey of a French preposition
263
In this example joy is described as lasting 3 days, where the temporal DP 3 days specifies not only the duration of joy but, by the same token, its intensity. Here we can see the influence of the co-text on the semantic interpretation: temporal items (3 days) associated with scalar predicates ( joy) give rise to degree readings. The same goes for (33) in which jusques au cuer crever specifies not only the terminal point of loving but also provides the intensity of love. Thus, a temporal DP occurring with a scalar predicate gives rise to a scalar reading: (33)
Sire , fet ele , por ce que , se vos m’ amiez jusques au cuer crever , ne porriez vos a moi avenir por nule chose , car ge aim un chevalier vers qui ge ne fausseroie por nule riens del monde (Anonymous, Mort le Roi Artu, 1230, p. 024) ‘My Lord, she said, because even if you loved me until the heart breaks (so much that your heart would break), you would not be able to approach me (obtain my love) by any means, for I love a knight whom I wouldn’t betray for anything in the world’
3.2.4. Spatial þ scalar. Another type of hybrid scalar reading arises from contiguity between spatial and degree readings (as in (34) and (35)): (34)
(35)
Ha! Perceval , fet li preudons , toz jorz seras tu nices ! Si ne conois mie cele damoisele qui te mena jusqu’ a pechie´ mortel , quant li signes de la Croiz t’ en delivra? (Anonymous, Queste del Saint Graal, 1220, p. 112) ‘Oh Perceval, said the worthy man, you will always be naı¨ve! Obviously, you don’t know this lady that led you to a mortal sin, when the sign of the cross delivered you from it.’ Mes toutes voies por la venjance de ce que tu estoies issuz de son servise te mena Nostre Sires jusques a paor de mort , por ce que tu te fiasses mielz une autre foiz en l’ aide Nostre Seignor que en ta force. (Anonymous, Queste del Saint Graal, 1220, p. 46) ‘But nevertheless as a revenge for no longer being in His service, our Lord led you to the fear of death, so next time you will trust in God’s assistance rather than in your own force.’
In (34), jusqu’ a pechie´ mortel is a complement of the verb of motion mener (‘lead to’), but apart from denoting a destination, it also specifies the extreme degree of seduction. This is due to the abstract nature of the complement (peche´ mortel). ‘Deadly sin’ is actually a noun denoting a resultant state of seduction or its culminating point, hence the idea of extreme degree. Another example of spatial reading combined with degree is frequent whenever the complement of jusqu’a` denotes a body part (examples (36)–(39)): (36)
Lancelos monte seur un destrier fort et isnel et couvert de fer jusqu’ en l’ongle del pie´. (Anonymous, Mort le Roi Artu, 1230, p. 194) ‘Lancelot gets up on his warhorse, strong and quick and covered with steel down to the hoof’
264
Silvia Adler and Maria Asnes
In (36), in addition to the spatial reading, the body part referring to extremities (ongle del pie´) emphasizes the extreme degree of the iron covering. The utilization of ‘hoofs’, a place where the horse is not supposed to be covered, triggers a scalar reading underlining the ultimate degree of being armed or covered. (37)
(38)
si li trenche le hiaume et la coife de fer et le fent jusqu’ es denz ; il estort son coup et le giete mort tout estendu a terre. (Anonymous, Mort le Roi Artu, 1230, p. 255) ‘Thus, he cleaves his helmet and his iron cap down to his (as far as) teeth; he twists his neck off and throws him dead on the ground’ Je vous conseille´ bien a droit selonc le siecle que l’ en voit , vous me feristes d’ un baston; sanglanz en fui jusqu’ au talon. (Anonymous, Roman de The`bes (vol. 2) verse 7778) ‘I advise you according to the rule of the world, you injured me with a stick, I was bleeding down to heel’
In (37) and (38), ‘teeth’ and ‘heel’ denote extremities to emphasize the intensity of the blow. (39)
Diex , trai t’ espee de ton fuerre , Ses porfent toz jusqu’ es entrailles. (Gautier de Coinci, Miracles de Nostre Dame (vol. 2), verse 1219) ‘For God’s sake, draw your sword from your sheath, strike him down to the entrails’
Entrailles (entrails) in (39) denotes not only the depth of the sword penetration but also the intensity of the injury. 3.2.5. Quantificational. As opposed to the general assumption according to which non spatiotemporal readings are subsequent derivations of the spatio-temporal one, the quantificational readings are already found in the beginning of the 13th century, at least in two texts: (40)
(41)
(42)
A ces paroles voient venir par mi la mestre rue jusqu’ a dis chevaliers armez qui lor dient que il se rendent ou il les occiront. (Anonymous, Queste del Saint Graal, 1220, p. 229) ‘Upon these words, they see coming by the main road up to ten armed knights that tell them to surrender or they will kill them’. Et lors a il encontre´ jusqu’ a vint homes armez qui portoient en une biere chevaleresse un home ocis novelement. (Anonymous, Queste del Saint Graal, 1220, p. 87) ‘And when he encountered up to twenty armed men that carried in a knightly bier a man that was recently slaughtered’. Lors s’en vont armer par laienz jusqu’ a quarante chevaliers, non mie par leur bone volente´, mes por ce qu’ il le couvient a fere , car li rois leur avoit commande´ de bouche. (Anonymous, Mort le Roi Artu, 1230, p. 120) ‘When up to forty knights go armed, not by their good will, but by that which they ought to do, since the king has commanded them from his mouth’.
A roots journey of a French preposition
(43)
(44)
265
Dame, vos avroiz encore assez greigneur respit que vos n’avez demande´ : il vous donront respit jusques a uit jours ; mes que vos me creantoiz que vos a celui terme feroiz outreement ce qu’ il vos requerront! (Anonymous, Mort le Roi Artu, 1230, p. 175) ‘My Lady, you will have much more grace than what you have asked for: they will give up to eight days of grace; provided you promise me that after that time you will do what they request’. Si vient au mestre huis dou pale´s, qui clos ert, si l’ uevre et descent en la cort aval ; si troeve chevaliers et escuiers jusqu’ a quinze qui l’atendoient et estoient venu avec lui. (Anonymous, Queste del Saint Graal, 1220, p. 8) ‘He comes to the main door of the palace, that is closed, opens it and comes down to the court; he finds there up to fifteen knights and squires that were waiting for him and came with him’.
These quantificational readings from Old French seem to function in a manner similar to their modern counterparts. For instance, it is possible to omit jusqu’a`, which functions here as a modifier of a numeral quantifier, without affecting the acceptability of the sentence. The entire phrase functions as a DP and not as a PP. However, as opposed to modern quantificational readings, Old French jusqu’a` can be used in neutral (pragmatically unmarked) assertive contexts, whereas the modern jusqu’a` suggests a discrepancy between the actual state of affairs and the speaker’s expectations (universe of beliefs) and is therefore frequently found in modal contexts (cf. (45)), its Old French counterpart simply provides an upper boundary of an approximate evaluation and encounters no obstacles in appearing within assertive neutral utterances (cf. examples (40)–(44) above). (45)
a. ??J’ai invite´ jusqu’a` 10 personnes (neutral assertive context) ??’I invited up to 10 people’ b. J’ai pu inviter jusqu’a` 10 personnes (alethic (capacity) or deontic (permission) possibility) ‘I could invite up to 10 people’ c. J’inviterai jusqu’a` 10 personnes (eventuality, future) ‘I will invite up to 10 people’ d. J’ai invite´ jusqu’a` 10 personnes en dix minutes (delimiting temporal expression) ‘I invited up to 10 people in ten minutes’ e. J’ai re´ussi a` inviter jusqu’a` 10 personnes en dix minutes (idea of achievement (re´ussir) and delimiting temporal expression) ‘I managed to invite up to 10 people in ten minutes’ f. On m’a dit/conseille´/permis d’inviter jusqu’a` 10 personnes (context of order/advice/ permission) ‘I was told to/ they allowed me to invite up to ten people’ g. Invitez jusqu’a` 10 personnes (deontic context of prescription) ‘Invite/you should invite up to 10 people’
266
Silvia Adler and Maria Asnes
The marked nature of the utterances in (45) results from the fact that the given quantity could have been lower. Although jusqu’a`, by its lexical meaning, imposes a limit, there is surprisingly no real limitation in all these modern contexts. On the contrary, jusqu’a` is used here to indicate a new superior15 value surpassing a pre-established standard. Let us notice that the sentences in question express possibility or capacity. Are we facing here a case of subjectification? This question will be dealt with in the next section, but in the meantime we will say that, as opposed to what has been shown by Visconti (2004, 2005a, b) regarding addirittura and Hansen (2005a, b) regarding enfin, jusqu’a` has not evolved into a discourse marker, but rather has preserved in Modern French a modal meaning, already present in its Latin ancestor usque.
4. SYNTACTIC EVOLUTION
AND
SEMANTIC DIVERSITY
4.1. Syntactic evolution It has been shown that, in Latin, usque was an adverb used as a specifier of the preposition ad. Gradually, from a free emphatic particle, it became a part of a complex preposition usque ad as a result of its fixed pre-prepositional position. In Old French, similarly to Latin, there was no clear-cut distinction between the categories of adverb and preposition (Buridant, 2000; Marchello-Nizia, 2002). Already then, the complex preposition jusqu’a` manifests paradigmatic ambiguity. Whereas in spatio-temporal and scalar contexts it behaves as a real, non-omissible preposition establishing a relation between two predicates, in the quantificational usage it functions more like a specifier modifying either a numeral quantifier or a DP. In this latter usage, the specifier is an omissible constituent, which does not establish any relation between two predicates (cf. Section 2.3). The same situation persists in Present Day French where for one linguistic item two syntactic paradigms – that of preposition and that of specifier – coexist at any given point of evolution. Thus, jusqu’a` exhibits properties of a polyfunctional morpheme. However, if we refer to the criterion of paradigmaticity studied by Lehmann (1995 [1982]), we notice that jusqu’a` as a specifier is not entirely paradigmatized since it does not have the same distributional properties as other adverbial specifiers, such as tout au plus, au moins. Although
15 Although (45b) could also be used in a context where the speaker had hoped to be able to invite more than 10 people, this would not be the most pragmatically efficient choice. If the speaker had wished to express a limitation, she would have rather used: pas plus que 10 personnes (‘no more than 10 people’) or ne pouvoir inviter que 10 personnes (‘be able to invite only 10 people’).
A roots journey of a French preposition
267
jusqu’a` – specifier – can be paraphrased by tout au plus, these two items do not share the same categorial status. For example, being an adverbial, tout au plus is mobile. This is not the case for jusqu’a` (cf. examples (12)–(15)). If jusqu’a`, as a specifier, is not entirely paradigmatized, this could mean that we do not have here a clear-cut evidence for grammaticalization, at least if we consider this criterion as an evidence – among other criteria – of grammaticalization. In the Latin example (46), usque ad already functions as a specifier: (46)
Et vidit ibi cum eis ipsum Guillemum Piscatorem et Ginhosam uxorem eius, et usque ad quindecim vel viginti personas, inter quas erat sicut credit Pontius vel Guillelmus Ramfredi, vel ambo. (Registre de l’inquisition de Toulouse 1273–1280, http:// jean.duvernoy.free.fr/text/pdf/Parnaclatin.pdf ) ‘And sees there the same Guillemum the fisherman and his spouse Ginhosam, and up to fifteen or twenty people, among which was, as believed, Pontius or Guillelmus Ramfredi, or both’.
4.2. Semantic diversity 4.2.1. Contiguity and metaphorization. Different scholars (see for instance Goyens and De Mulder, 2002: 192–193) following Traugott and Ko¨nig (1991) attribute semantic changes to contextual reinterpretation, that is to say to the fact that an expression appears in a new context. In our case it seems that the scalar reading emerged, among others, from a possible contiguity of jusqu’a` with temporal predicates. Thus, aimer jusqu’a` la mort can denote the temporal endpoint of love (‘until death do us part’) or, by a metaphorical extension, the degree of loving which does not necessarily have to last till death. The English translations disambiguate the denotation of aimer jusqu’a` la mort: ‘to love until death’ or ‘to love to death’. The first translation emphasizes the temporal endpoint while the second one stresses the intensity, although the first one may also convey the scalar reading, according to some speakers. Thus, in (47): (47)
a. He loved her to death ( ¼ the lovers are not dead, death is a metaphor for the ultimate degree of love) b. He loved her until death (.at least one of the lovers is dead now)
Hebrew possesses another strategy, related to the use of the definite article, in order to distinguish temporal and scalar readings: (48)
a. Leehov ad mavet ‘To love to death’ b. Leehov ad ha-mavet ‘To love to the death’
268
Silvia Adler and Maria Asnes
In (48a) the noun mavet (‘death’) is used without an article, intensionally, and, as such, it can only function as an intensifier. On the other hand, in (48b) where this noun is preceded by the definite article, and thus used extensionally, only a temporal reading is available. The same goes for spatial readings which can metaphorically be reinterpreted as scalar. For example, arme´ jusqu’aux dents/ongles (‘armed to teeth/nails’) or habille´ jusqu’au bout des ongles (‘dressed up from head to toe’), spatial in origin, are reinterpreted as scalar in the sense that the body parts used in these expressions to denote extremities can convey the meaning of the culminating point or completeness as well. The only scalar readings we have found in Old and Middle French are hybrid, in the sense that they are metaphorical extensions of spatial and temporal interpretations. As opposed to Old and Middle French, in Modern French we find pure scalar readings, where the complement of the preposition is not spatio-temporal in its nature ((49) and (50)): (49)
(50)
L’ autre, laı¨c, chevalier, pe´lerin, croise´, roi ceint de la premie`re couronne chre´tienne, brave jusqu’ a` la te´me´rite´, n’ he´sitant pas plus a` exposer sa vie qu’ a` courber sa teˆte devant *Dieu (Montalembert Ch./Histoire de Ste. Elisabeth/1836, p. LXXX/Introduction) ‘The other laic, knight, pilgrim, baptized, king encircled with the first Christian crown, brave up to temerity, not hesitating either to expose his life or to bend his head in front of God’ Tu sais comme il e´tait courageux jusqu’ a` l’imprudence. (P847/Pasteur. L/ Correspondance T.3/1884, p. 391) ‘You know to what extent he was courageous up to imprudence’
Since the purely scalar readings seem to be posterior to hybrid ones, one could claim for a semantic shift through metaphorization (Traugott and Dasher, 2002). Visconti (2005a: 246) claims with respect to the expression perfino alla morte that ‘the scanning of all points/instants to the last one on a space/time line is projected onto the mental scanning of a scale of values to mark its endpoint’.16 This would be the case here as well. 4.2.2. Subjectification17. Subjectification consists in encoding the speaker’s attitude or appraisals. This section addresses the question of whether the development of the scalar and the quantificational usages of jusqu’a` reflects a diachronic process of subjectification. Visconti (2005a) argues that the scalar value of perfino and addirittura is a result of a conventionalization of an evaluative component originally constructed by context and gradually
16 Following Traugott (2001), Visconti (2005a) associates the development of the scalar meaning to the presence of a directional component in the original meaning of a linguistic item. 17 Langacker (1990), Stein and Wright (1995), Traugott (1995), Traugott and Dasher (2002), Hopper and Traugott (1993), Visconti (2004, 2005a, b).
A roots journey of a French preposition
269
absorbed by these lexemes. In other words, these lexemes underwent a process of subjectification, for the speaker’s attitude ended by being grammatically codified. Do the scalar and the quantificational values of French jusqu’a` manifest subjectification? With respect to the scalar value of jusqu’a`, we have outlined (2.2) a transition from a hybrid scalar value (temporal or spatial items associated with scalar predicates give rise to a scalar reading) to a ‘pure’ scalar value in Present Day French. In that sense, one could argue for subjectification, for we have no longer a situation where context has an influence on interpretation. In other words, Present Day jusqu’a` does not need any spatial or temporal components in its scope in order to denote scalarity. Yet, there exist two arguments challenging subjectification. First, jusque in itself – and not the PP headed by jusque – originates in a Latin particle denoting reinforcement (2.1) and functioning as a degree modifier. In other words, the speaker’s attitude has been encoded already in Latin usque. This fact suggests subjectivity rather than subjectification (consisting in a dynamic evolution). Second, as far as the syntactic correlates of subjectification are concerned, we do not observe here a case of syntactic expandability or increased freedom. Jusqu’a` remains in its Present Day French scalar usage truth – conditionally non-redundant and syntactically constrained (the preposition establishes a relation between two predicates). What about the quantificational value? Syntactically speaking, in the quantificational reading illustrated by examples such as (51), although jusqu’a` has only one predicate in its scope, it is not mobile. Moreover, its potential omission does not suggest redundancy: while having no influence on the syntactic acceptability, its omission nevertheless affects the truth conditions of the clause. (51)
a. Inviter ‘Invite b. Inviter ‘Invite
jusqu’a` trois personnes up to 3 people’ trois personnes 3 people’
Example (51b), where jusqu’a` is omitted, is syntactically acceptable but its truth conditions are different from those of example (51a): (51u) a. INVITER (x)4PERSONNE (x)4 7PERSONNE7r3 (51u) b. INVITER (x)4PERSONNE (x)4 7PERSONNE7 ¼ 3 These facts discredit the idea of subjectification, but an important question remains to be answered regarding the fact that the Modern quantificational value, where jusqu’a` is a specifier of a quantifier, suggests a discrepancy between the actual state of affairs and the speaker’s universe of beliefs and therefore is favored in modal contexts. This situation is different from that observed in Old French, where jusqu’a` is used in neutral (pragmatically unmarked) assertive contexts. For instance: (52)
Il y a jusqu’a` dix trains par heure entre l’ae´roport de Zurich et la gare centrale
270
Silvia Adler and Maria Asnes
‘There are up to (as many as) ten trains per hour between Zurich’s airport and the central station’ In this example the railway company presents as an achievement the number of trains between Zurich’s airport and the central station, which surpasses the expectations of the passengers. Is that a case of subjectification? Not necessarily. Rather, this seems to be a scalar implicature: although jusqu’a` imposes a right (upper) boundary, its choice suggests that, as far as the speaker knows, the quantity could have been lower ( jusqu’a` DP is presented as a culminating point of a path containing lower quantities and not as a limitation).
5. CONCLUSION Jusqu’a` seems to be a polyfunctional form which did not evolve much since usque shifted from an adverb into a prepositional specifier. The emphatic element claimed to be present in the scalar and the quantificational usages is in fact present even in the spatial usage, in the sense that, contrary to ad (a`) alone, jusqu’a` is a directional preposition denoting also a path. This property of directionality and path is a common denominator for all other readings shared by this preposition. We thus conclude that in our case there is no real evolution from spatial to non-spatial, concrete to abstract, but rather that one sense core, in itself abstract, is applicable to various domains, such as space, time, scalarity. All of these conserve the idea of axis, continuum, and oriented scale.18
ACKNOWLEDGMENT Many thanks to our colleagues Cyril Aslanov and Isabelle Martin for their helpful comments on our translations. All possible errors are solely ours.
REFERENCES Adler, S. and M. Asnes (2004). ‘‘Les comple´ments de degre´ en jusqu’a`’’. Travaux de linguistique 49: 131–157.
18
Thus we join Fagard (2002) and Selosse (2002) who claim that when different meanings are found in the earliest texts, this discredits the primacy of the spatial reading.
A roots journey of a French preposition
271
Adler, S. and M. Asnes (2010a). ‘‘A la recherche des quantifieurs perdus’’, Actes du colloque ‘‘La quantification et ses domaines’’, 19–21 octobre 2006, Universite´ Marc Bloch, Strasbourg, Presses Universitaires de Caen. Adler, S. and M. Asnes (2010b). Pre´positions au service de la scalarite´. Geneva: Droz. Bach, E. (1986). ‘‘The algebra of events’’. Linguistics and Philosophy 9: 5–16. Buridant, C. (2000). Grammaire nouvelle de l’ancien franc- ais. Paris: SEDES. Casselman, B. (1995). Canadian words: A comic browse by Bill Casselman through words and folk sayings invented by Canadians. McArthur and Company http://www.billcasselman.com/casselmans_ canadian_words/ccw_seven.htm Dobrovie-Sorin, C. and C. Beyssade (2004). De´finir les inde´finis. Paris: CNRS Editions. Fagard, B. (2002). ‘‘Evolution se´mantique des pre´positions spatiales de l’ancien au moyen franc- ais’’. Lingvisticae Investigationes 25 (2): 311–338. Goyens, M. and W. De Mulder (eds.) (2002). ‘‘Grammaticalisation: Le cas des pre´positions locatives.’’ Special issue of Lingvisticæ Investigationes 25:2. Hansen, M.-B. M. (2005). ‘‘A comparative study of the semantics and pragmatics of enfin and finalement, in synchrony and diachony’’. French Language Studies 15: 153–171. Hansen, M.-B. M. (2005). ‘‘From prepositional phrase to hesitation marker: The semantic and pragmatic evolution of French enfin’’. Journal of Historical Pragmatics 6 (1): 37–68. Hopper, P. J. and E. C. Traugott (1993). Grammaticalization. Cambridge: Cambridge University Press. Ilinski, K. (2003). La pre´position et son re´gime: E´tude des cas atypiques. Paris: Champion. Krifka, M. (1999). ‘‘At least some determiners aren’t determiners’’, in K. Turner (ed.), The semantics/ pragmatics interface from different points of view. Oxford: Elsevier, 257–291. Langacker, R. W. (1990). ‘‘Subjectification’’. Cognitive Linguistics 1: 5–38. Lehmann, C. (1995) [1982]. Thoughts on grammaticalization. Mu¨nchen: Lincom Europa. Marchello-Nizia, C. (2002). ‘‘Pre´positions franc- aises en diachronie: Une cate´gorie en question’’. Lingvisticae investigationes 25 (2): 205–221. Melis, L. (2001). ‘‘La pre´position est-elle toujours la teˆte d’un groupe pre´positionnel?’’. Travaux de linguistique 42–43: 11–22. Selosse, P. (2002). ‘‘Pour une phytothe´raphie de la polyse´mie pre´positionnelle’’. Lingvisticae investigationes 25 (2): 339–359. Stein, D. and S. Wright (eds.) (1995). Subjectivity and subjectivisation: Linguistic perspectives. Cambridge: Cambridge University Press. Traugott, E. C. (1995). ‘‘Subjectification in grammaticalisation’’, in D. Stein and S. Wright (eds.), Subjectivity and subjectivisation: Linguistic perspectives. Cambridge: Cambridge University Press, 31–54. Traugott, E. C. (2001). ‘‘Legitimate counterexamples to unidirectionality’’. Paper presented at Freiburg University, October 17. Traugott, E. C. and R. B. Dasher (2002). Regularity in semantic change. Cambridge: Cambridge University Press. Traugott, E. C. and E. Ko¨nig (1991). ‘‘The semantics-pragmatics of grammaticalization revisited’’, in E. C. Traugott and B. Heine (eds.), Approaches to grammaticalization Vol. I. Amsterdam: John Benjamins, 198–218. Vendler, Z. (1967). ‘‘Verbs and times’’, linguistics in philosophyIthaca, NY: Cornell University Press. 97–121 Vincent, N. (1999). ‘‘The evolution of c-structure: Prepositions and PPs from Indo-European to Romance’’. Linguistics 37 (6): 1111–1153. http://www.essex.ac.uk/linguistics/LFG/www-lfg.stanford.edu/pubs/papers/ vincent/vincent-1998-0525.ps
272
Silvia Adler and Maria Asnes
Visconti, J. (2004). ‘‘Sintassi e uso delle particelle perfino, pesino e addirittura in italiano antico’’, in M. Dardano and G. Frenguelli (eds.), SintAnt: La sintassi dell’italiano antico. Roma: L’Aracne, 445–463. Visconti, J. (2005). ‘‘On the origins of scalar particles in Italian’’. Journal of Historical Pragmatics 6 (2): 237–261. Visconti, J. (2005b). ‘‘From preposition to specifier: Subjectivation vs grammaticalization in the evolution of Italian scalar particles’’. Paper presented at the conference ‘‘From Ideational to Interpersonal: Perspectives from Grammaticalization’’, Leuven 10–12 February.
DICTIONARIES Dictionnaire de l’ancienne langue franc- aise et de tous ses dialectes du IXe au XVe sie`cle, tome IV, Godefroy F. Dictionnaire e´tymologique de la langue franc- aise. Bloch, O. et von Wartburg, W. Paris: Presses Universitaires de France, 1975. Dictionnaire e´tymologique et historique. Dauzat, A. et al. 1974. Paris: Larousse. Dictionnaire de la langue franc- aise du XVIe sie`cle, Huguet, E., 1932–1949, Paris: Librairie Ancienne Honore´ Champion. Tre´sor de la Langue Franc- aise (TLF) informatise´, http://atilf.atilf.fr/tlf.htm
Corpora Base du franc- ais me´divie´val (BFM) electronic corpus, http://weblex.ens-lsh.fr/wlx/cgi/weblex-f? corpus ¼ bfm. Compiled by the Laboratoire Analyses de Corpus de l’E.N.S. Lettres et Sciences Humaines, Lyon. Frantext electronic corpus. http://atilf.atilf.fr/frantext.htm The Song of Roland, translated, with an Introduction by W. S. Merwin, 2001, New York: The Modern Library.
Current Trends in Diachronic Semantics and Pragmatics Edited by Maj-Britt Mosegaard Hansen and Jacqueline Visconti r 2009 by Emerald Group Publishing Limited. All rights reserved.
14 THE GRAMMATICALIZATION OF PRIVATIVE ADJECTIVES: THE CASE OF MERE Elke Gehweiler
1. INTRODUCTION Present Day English (PDE) mere is a negative intensifier, or downtoner in the sense of Quirk et al. (1985: 430, 590), having ‘‘a lowering effect, usually scaling downwards from an assumed norm’’. Mere developed from a privative adjective meaning ‘pure, unmixed’ through a process of grammaticalization and subjectification. Privative adjectives are adjectives which denote or predicate privation, i.e. adjectives whose meaning is characterized by the absence of a certain quality or attribute (cf. also Leisi, 1967). Other examples include bare ‘uncovered’, German bloX ‘uncovered’, or the now obsolete German adjective lauter ‘pure, unmixed’. Privative adjectives come to be grammaticalized with a more subjective meaning frequently. In this paper I present a case study for mere. First, the present day adjectival status of mere is discussed, showing that PDE mere is a peripheral adjective and may be regarded as grammaticalized. Then a short corpus analysis of mere is presented with data from the British National Corpus (BNC), showing that mere is used in only a small number of different patterns in PDE. The second part of the paper describes the diachronic situation for mere. Based on the analysis of two diachronic corpora, the Oxford English Dictionary (OED) quotations database and the Parsed Corpus of Early English Correspondence (PCEEC),1 I will trace the emergence of
1
Here I would like to thank Arja Nurmi who made this corpus available for me.
274
Elke Gehweiler
today’s intensifier mere from its oldest meaning ‘pure, unmixed’ via the metaphorical meaning ‘morally good’.
2. SYNCHRONIC
MERE
2.1. The categorial status of PDE mere In grammars of English, members of the category adjective are defined with recourse to the following properties (cf. e.g. Quirk et al., 1985; Biber et al., 1999; Huddleston and Pullum, 2005) where central members of the word class display all of the properties: (i) use in attributive position, i.e. as modifiers of nouns, (ii) predicative use, i.e. as complements, (iii) gradability, meaning that they can be modified by degree adverbs and (iv) degree of comparison, i.e. comparative and superlative uses. It will be shown that in PDE mere is not a central adjective according to these criteria. 2.1.1. Attributive use. In fact, mere only fully fulfils the first of the four defining criteria for adjectives. The corpus examples show that mere is exclusively used in attributive position as modifier of nouns. But in its attributive use, too, mere is quite restricted: Firstly, it occurs in a limited set of patterns. These uses will be analyzed in detail in Section 2.3. Moreover, when there is more than one adjectival modifier within the premodifying string, mere is always first. In the BNC there is only one occurrence where mere is preceded by another adjective. (1)
The conflicts and inconsistencies present in our commonsense and other mere primitive concepts of inquiry are resolved in a procedure which exhibited a harmony between presumption, aim, and method (BNC 2)
The item other, however, is a word whose categorical status is equivocal. It has both determiner-like and adjective-like properties. In examples as the above other is part of the determiner-unit: it helps the hearer to establish which instances the speaker is referring to, indicating that the instances referred to are not identical with other discourse referents (cf. Breban, 2006: 147ff for a detailed analysis of other). Items like other have therefore been called ‘postdeterminers’ (cf. Sinclair, 1991: 70; Halliday, 1985: 162; Quirk et al., 1985: 262) or ‘semideterminers’ (cf. Biber et al., 1999: 280). As part of the determiner-unit, postdeterminers immediately follow the (central) determiner in the NP and precede all (other) modifiers. If other is regarded as part of the determiner-unit, then mere in the above example is not preceded by another modifier, but is itself the only modifier within the NP.
The grammaticalization of privative adjectives: the case of mere
275
On the other hand, uses where mere precedes other adjectives are quite frequent: more than 300 in the BNC, accounting for around 10 per cent of all occurrences of mere. 2.1.2. Predicative use. There are no predicative uses attested in the whole BNC.2 However, the OED gives two current examples where mere is used in that way and ascribes predicative mere the meaning ‘insignificant, ordinary; inadequate, feeble’ (OED, s.v. mere, a.2). (2)
(3)
In such situations American and Dominion soldiers have always felt less ‘mere’ than the English. (OED: 1955, s.v. mere, a.2) Such programmes are sometimes dismissed as being merely ‘slick’, when slickness is anything but mere. (OED: 1989, s.v. mere, a.2)
A search in the GoogleGroups newsgroups3 revealed a further 11 different predicative uses of mere.4 In 10 of these uses, as well as in the second occurrence from the OED, mere refers back in the context and takes up a previous use of mere or of the adverb merely in the conversation. In 9 of these 10 uses from the GoogleGroups quotation marks or other metalinguistic marks (like*) signal that the speakers perceive this use of mere as exceptional and regard it as a kind of quotation, i.e. what we have here is a metalinguistic use of the form. For the first example from the OED, given in (2), where inverted commas also function as metalinguistic signal for exceptional use, the context is not large enough to be sure, but here, too, mere seems to similarly refer back to a previous use of mere or merely. (4) A: Likewise, MDs are looked down upon in biomedical research until they have proven themselves, for good reason. MDs are not (in general) trained to do any part of research, from formulating the questions to doing basic lab tasks like diluting stocks to analyzing the data. Out of the environment they’ve been trained for, they’re merely MDs. [ . . . ] B: So, which doctorates are ‘‘mere’’ and which are worthy of the title ‘‘Dr.’’? How does one distinguish between them? (GoogleGroups: Gruppe: soc.motss, April 22, 1998) (5) A: It still does not make it right that mere property is regarded as more valuable than life. B: Not all property is ‘‘mere,’’ and not all life has value. (GoogleGroups: Gruppe: alt.politics.british, July 29, 2003)
2
Actually there are three examples where mere is used predicatively. These are, however, from a linguistics textbook and are marked as ungrammatical. 3 Searched on 23 March 2007 for ‘‘is mere and’’, ‘‘are mere and’’, ‘‘was mere and’’, ‘‘were mere and’’. 4 Two poetic uses, where the predicative use of mere was also due to rhythmic reasons, were dismissed.
276
Elke Gehweiler
There is one example that could be classified as ‘‘real’’ predicative use of mere, where mere does not refer back. It is however most likely written by a non-native speaker of English and is immediately commented on by the other writer, who seems to perceive it as wrong. (6) A: The problem is mere and cant give place for any comment. However this quite normal fact is presented as an anomaly of the Newtonian, or Galilean physics. B: What does it mean for a problem to be ‘‘mere’’? (GoogleGroups: Gruppe: sci.physics.relativity, December 18, 1998) Summarizing it can be said that PDE mere does not fulfil the second criterion for central adjectives, i.e. it has no established predicative use. 2.1.3. Gradability. Furthermore, mere is not gradable (*a very mere mortal). A search in the BNC for ADV þ mere did not reveal any occurrences where it is modified by a degree adverb. 2.1.4. Degree of comparison. There are no comparative uses of mere attested in the BNC (*a merer mortal), but there are a number of superlative uses, i.e. 123 in the whole BNC. Compared to about 3300 occurrences of the positive form mere this yields a ratio of 1:27 of superlative versus positive uses. Such a ratio is not unusual; central adjectives such as small and old, for example have a ratio of 1:40 and 1:36 respectively; good has 1:2.3. s. Inside , it felt so phoney that nd when we begin to gain , if only -sounding ‘ departed ’ . Yet it is then dawning recognition . It took or the most fleeting expression ,
the the the the the
merest merest merest merest merest
glimmer of amusement would have sent glimpse of the deeply ingrained bruta hint of hope in a story of otherwise instant , but as a process it never c muscular twitch , rise unbidden to an
e prints that they can send off at . Despite a conspicuous absence of started hitting the ball around , t 2’07 ‘ , withdrawing his tone to ping water and a palm tree. It was
the the the the the
merest merest merest merest merest
rumour of work . You will need to sen shred of evidence that Nizan had ‘‘ be tap sending it miles across a field a thread of sound , barely grazing the trickle , four cubic metres a day acc
The superlative uses of mere, however, are quite restricted, as the concordance lines above show. The superlatives occur virtually exclusively – like most superlatives – with the definite determiner the, but they modify in most examples a small size noun (SSN) or another noun containing a meaning element ‘not much’, in about half of the examples within a partitive construction. SSNs are nouns that denote a small amount, small size, low volume etc. SSN constructions, which normally consist of ‘‘a SSN, mostly preceded by an (indefinite) determiner and sometimes (adjectival) premodifiers, and typically followed by of and a second nominal’’ (Brems, 2007: 293f ), can come to be grammaticalized (cf. also Traugott, 2007). During this grammaticalization process the SSN construction with the original structure [SSN] þ [of þ N2], with the SSN having lexical head noun status, as in a scrap of paper, is re-bracketed as [SSN þ of ] þ [N2] and reinterpreted as a quantifier,
The grammaticalization of privative adjectives: the case of mere
277
as in (not) a scrap of evidence, N2 being the head noun. These grammaticalized constructions combine with a larger range of N2 collocates (cf. Brems, 2007; Traugott, 2007). Many of the SSN constructions in which merest is used are grammaticalized, too, the merest hint of being the most common of them (cf. also glimmer, shred, thread in the concordance lines above). The definite article in these uses is triggered by the superlative form. 11 frock coat and waistcoat , with oup so far , and if there was even nd bordered in a silk plaid to add ble representation of the nation , -grey cloud , now empty of all but
the the the the the
merest merest merest merest merest
hint of a beard on his face , as he r scintilla of justice in this doomed p touch of bright colour . Ruched light whiff of criticism draws the fire of wisps of white . On every side of the
Premodifiers in SSN constructions, like merest, ‘‘do not intrude upon the internal cohesion of the grammaticalized SSN structure, but rather reinforce or echo it’’ (Brems, 2007: 302). Brems shows that grammaticalized SSN constructions often function as downtoners, for example whiff of with which merest also occurs. It is therefore not surprising that merest – remember that mere itself is a downtoner, too – is restricted to uses with such downtoning SSNs, adding further negative intensification to the SSN construction. 2.2. Evidence for grammaticalization and subjectification Although mere used to be a central member of the word class adjective at an earlier stage of the language (see Section 3), in PDE it is subject to severe grammatical restrictions and is classified as peripheral adjective according to the four defining properties for adjectives: it is not used predicatively, it is not gradable, there are no comparative uses attested, and its superlatives predominantly occur in the pattern [the merest SSN (of N)]. During the development from central to peripheral adjective, the meaning of mere has shifted from the referential meaning ‘pure, unmixed’ to a (negative) intensifying meaning, which the OED paraphrases as ‘Having no greater extent, range, value, power, or importance than the designation implies; that is barely or only what it is said to be’ (s.v. mere, a.2). It can be said to have a lowering effect, scaling downwards from an assumed norm and can therefore be called a ‘downtoner’ with Quirk et al. (1985:430, 590). The present day meaning of mere is obviously more subjective than its original meaning. Furthermore it does no longer have referential meaning. Adamson (2000) has proposed a correlation between subjectification and position within the NP. She has shown that as adjectives change their meaning historically, and become more subjective, they shift to the left within the premodifying string, i.e. when there is more than one modifier, the most subjective one comes first. This is also the case for mere: as shown above, when there is more than one adjectival modifier mere, is always first. The development of mere and the grammatical restriction it involves, it is argued, is also a case of grammaticalization, as it involves a shift from the centre to the periphery of the category
278
Elke Gehweiler
‘adjective’. Such decategorialization is typical of grammaticalization (cf. e.g. Hopper and Traugott, 2003: 106; Brinton and Traugott, 2005: 25). The fact that mere now always occurs as the first modifier in the NP suggests a proximity to or a tendency towards the word class determiner. Denison (2006) has shown that adjective and determiner are adjacent categories with fuzzy boundaries and that elements can move from adjective to determiner (cf. also Breban, 2006). This analysis is supported by the analogous development of the German privative adjective lauter ‘pure, unmixed’, which first turned into an intensifier, synonymous to PDE mere, but developed further into a determiner whose meaning can be paraphrased with ‘(too) many, only’, i.e. it clearly underwent grammaticalization. The micro-processes that led to the emergence of the downtoner mere will be described in Section 3.3. 2.3. Corpus analysis of PDE mere For the synchronic analysis of mere a random sample of 150 corpus occurrences of mere was drawn from the BNC. These examples were analyzed syntactically and semantically, showing that in PDE mere is used in a limited number of patterns. Minor deviations from the patterns do occur, but will not be discussed in more detail here; for a detailed corpus semantic analysis of mere cf. Gehweiler (2009). 2.3.1. [a þ mere þ QUANT þ MEASURE UNIT N]. In 30 uses mere occurs within an NP expressing quantification. The head noun is a noun denoting a unit of measure (such as minute, mile or per cent) or another countable noun. In these uses mere is always preceded by the indefinite article and followed by a cardinal numeral. Its meaning can be paraphrased with ‘only’: mere indicates that the quantity denoted by the NP is perceived as small by the speaker compared to some other quantity given explicitly or implicitly in the context. (7) (8)
Tootal bought Sandhurst in 1986 for d21.7 m, and sold it in July for a mere d4.38 m. (BNC A1S) inflation had fallen from a peak of 22 per cent in early 1980 to a mere 5 per cent (BNC A66)
2.3.2. [the þ mere þ N þ of þ NP]. Twenty of the 150 uses are instances of a binomial construction with of, which is introduced by the definite determiner the. The head nouns of mere all take a complement, which is expressed by the of-phrase. The binomial construction containing mere gives the cause for certain consequences (e.g. actions, attitudes, emotions) that are described (mostly) later in the same sentence. The function of mere in these uses is to indicate that the cause that leads to the described consequences is comparatively minor or trivial, even inappropriate, implying that one would usually expect something more serious to trigger the described
The grammaticalization of privative adjectives: the case of mere
279
consequences. Through pointing to the dissonance between cause and consequence, mere implicitly establishes a contrast or comparison (between actual and expected cause). In (9), for example the cause ‘‘having a man play a woman’s role’’ has the intention ‘‘to provoke laughter’’. By the use of mere, the speaker indicates that he/she regards this cause as inappropriate, i.e. he/she signals by the use of mere that in his/her view the fact that a man plays a woman’s role should not be intended to provoke laughter. (9)
(10) (11)
(12)
But there are some works in a lighter vein this one being a prime example for which one wonders if the mere fact of having a man play a woman’s role was intended to provoke laughter. (BNC J1A) The mere signal of takeover intentions normally engenders a rise in the share price (BNC EX2) Truly there must be something strange about the seaside, when the mere sight of seeing an ankle had made his heart beat the faster. How dear Maisie, or Natalia, and certainly Emma, would laugh to see him enslaved by an ankle (BNC H8A) Spencer lifted his arms in a flamboyant gesture and Emily felt physically sick at the mere thought of marrying him. (BNC CKD)
2.3.3. Contrasting patterns. In 94 of the remaining 100 corpus occurrences (the remaining 6 uses will not be discussed further here), mere is also used to establish contrast. Its function is to denote the ‘lower-ranking’ or ‘inferior’ of two items of the same kind that are being compared. In most cases another linguistic expression is present that helps to bring about a contrastive interpretation: Contrast is established by the use of, for example certain adverbials, prepositions, adjectives, subordinators or coordinators with a contrasting function, verbs that denote grading or contrast, relational nouns, or certain syntactic constructions. In a number of cases a contrastive interpretation arises only in a larger textual context. In the following some of the more frequent of these contrasting patterns will be discussed in more detail. The first pattern, [more þ than þ (a) þ mere þ N], is related to the comparative form of adjectives. It can be paraphrased as ‘not N (but)’. The pattern establishes contrast by referring to an assumption that is rejected and replacing it by another. (13)
Indeed, the relationship has been more than mere friendship, but one of familial depth. (BNC A0P)
The patterns [GO þ beyond þ (a/the) þ mere þ N] and [over and above þ (the) þ mere þ N] both involve prepositions that set two items in contrast. As before, mere is part of the NP that refers to the inferior of the contrasted items.
280
Elke Gehweiler
(14)
(15)
looking for some general structure for the undergraduate curriculum which goes beyond a mere aggregation of subjects (BNC FA3) Here, the relationships between managers and workers were tolerant and friendly, and involved loyalties and commitments over and above the mere exchange of wages for work. (BNC G0U)
In a further pattern, [((a/no) þ mere þ N)NEG þ (,) þ but], mere is again part of a NP whose referent is set in contrast to another item, here by the use of a coordinative structure involving but. The first assumption is rejected, i.e. negated, and replaced by another one. As in the previous uses, mere is part of the inferior part of the comparison, i.e. of the NP that is rejected. (16)
In these ways, they force their parents to recognise that they are no longer a mere extension of them or the family, but independent human beings with views and ambitions of their own. (BNC BLW)
In many uses the head noun of mere helps to bring about a contrastive reading. The nouns involved are relational nouns, such as symptom in (17), or fragment, or nouns like ripple in (18), child in (19), button, or whisper that denote something small. Relational nouns denote items that are a part of or in a certain way related to or dependent on something larger, superordinate or superior, i.e. they are dependent in their interpretation on something else and can therefore be said to contain a contrastive meaning element. For the second type of nouns, nouns that contain the meaning element ‘small’, it is exactly this meaning element – together with mere – that brings about a contrast-reading in that it implicitly establishes a comparison with something larger. (17)
(18)
(19)
he would be able to dispel the habits and characteristics of which these particular tastes were a mere symptom (BNC A7C) Mistakes are bound to happen sometimes. Regard them as an annoying incident, a mere ripple on the pool – do not drown in it, it is not the end of the world. (BNC C9R) The proposal to transfer Chinon, Loudun and Mirebeau to a mere child was surely just a trick giving Henry II an excuse to keep these three important castles in his own hands for many years to come. (BNC EFV)
In some uses of the sample it is the verb that establishes contrast. The examples include verbs that denote grading and thereby contain a comparative element, such as reduce (cf. (20)), waste
The grammaticalization of privative adjectives: the case of mere
281
away, or degenerate – all expressing comparison in relation to a lower degree – as well as verbs containing an element of contrast, such as replace in (21), suffice, compare or distinguish. As in the above uses, mere is part of the NP denoting the inferior of the contrasted elements. (20)
(21)
There are three grounds for this pervasive professional opposition to the system. First, ‘‘payment by results’’ demeaned education and reduced it to a mere cramming exercise in which all that is expected of pupils is a capacity for recall. (BNC CN5) But, as the battle for emotional security and economic prosperity starts to be won, questions of self-expression, esteem and self-realisation begin to replace mere selfdefence. (BNC HRM)
Sometimes a comparison interpretation arises through the syntactic construction in which mere is used, in (22) through the use of a copula construction. (22)
His flashes of brilliance make one want to leap up and cheer: ‘‘No English man or woman could play Antony and Cleopatra: passion and love are quite beyond them, they could never discard their suburban subservience. Antony and Cleopatra would always be a mere John and Norma.’’ (BNC CAL)
The analysis of 150 random occurrences shows that mere is mostly used within patterns that compare something inferior (e.g. subordinate, smaller, lower quantity) to something related that is superior to it (e.g. superordinate, larger, larger quantity). As a downtoner mere is always part of the NP denoting the lower, inferior of the items compared, i.e. it has a lowering effect, scaling downwards. Note that ‘inferior’ and ‘superior’ do not necessarily mean evaluatively inferior or superior: In most cases the relation between the two items that are compared is one of ‘objective’ superiority/inferiority. This is confirmed by the fact that mere rarely co-occurs with (evaluatively) negative head nouns.5 The restriction of mere to contrastive patterns follows from its meaning as a downtoner that is scaling downwards. Quirk et al. (1985: 590) note that ‘‘[s]uch scaling requires that the item or unit to which the intensifier applies is gradable’’, and indeed it is gradability in a wide sense (or rather ‘contrastability’) that unites all uses of mere.
5
Therefore the meaning of mere should not be called depreciative (cf. Wierzbicka, 1986).
282
Elke Gehweiler
3. DIACHRONIC
MERE
3.1. The meanings of mere At an earlier stage of the language mere used to be a central adjective with a clear referential meaning. In its oldest meaning, attested from Old English to Early Modern English (EModE), it is restricted to co-occurrence with mass nouns denoting substances, and its meaning can be paraphrased as ‘pure, unmixed’ (cf. (23), (26) and (27)). Later, mere came to be used metaphorically, co-occurring with nouns referring to human (and spiritual) beings and their inner states, i.e. their feelings, behaviour, or character (cf. (24) and (25)). Depending on the respective noun its meanings range from ‘honest, sincere’ to ‘virtuous, virginal’. These different shades of meaning of mere when co-occurring with nouns denoting human beings or inner states will be subsumed under the general semantic paraphrase ‘morally good’. The metaphorical meaning of mere is attested from Middle English (ME) to EModE. Furthermore mere had a number of other, related adjectival meanings, which did not play a role in the emergence of the PDE intensifier mere. As all meanings of mere can be paraphrased by using the adjective pure, which is itself highly polysemous, I will refrain in what follows from using ‘pure’ as a semantic paraphrase for mere altogether. Instead, the paraphrases of mere that are used (‘unmixed’ and ‘morally good’) are based on the semantic classes of the nouns with which it cooccurs (nouns denoting substances vs. nouns denoting human beings and their inner states). As a central adjective, mere ‘unmixed’ was used attributively as well as predicatively. When it was combined with other adjectives in its attributive use, it was not always the first modifier in the NP, as in (24), where it is preceded by the adjective true. (23)
(24) (25) (26)
(27)
Because a halfpeny cannot be made of such finenesse to beare any conuenient bulke, an other small peece shall also be coyned of three farthynges..whiche..shalbe of meere fine starling syluer. (OED: 1561, s.v. three- farthings) The true, mere, and sincere word of God. (OED: 1535, s.v. mere, a.2) He..dooth answer againe, by cauilling sophistication, & by meere affection. (OED: 1553–87, s.v. affection, n.) Willam þ te spicer & geffray . of hencsei þ tat þ to were . Portreuen & nicole . of kingestone þ tat was mere. (Helsinki: 1325, cmrobglo) (Kingstone: angel fish, monk fish) Our wine is here mingled with water and with myrrhe, there it is mere and unmixt. (OED: 1660, s.v. mere, a.2)
The OED quotations database contains examples where mere is modified by the degree modifier very, though only for the meaning ‘insignificant, ordinary; inadequate, feeble’ (cf. OED,
The grammaticalization of privative adjectives: the case of mere
283
s.v. mere, a.2), but it may well be assumed that both the adjectives mere ‘unmixed’ and mere ‘morally good’ were gradable. As discussed above, modification by adverbs of degree is no longer possible today. (28)
(29)
To give the choyce of a thousand thankes for every gewegawe; and sumtymes tooe for very meere Nifilles as it were only pro forma tantum. (OED: 1573–80, s.v. pro, n.1, prep., and a.) (nifle: a trifle; a thing of little or no value) She is still very mere but not quite so mere – in the strict sense of the word – as she was four weeks ago. (OED: 1893, s.v. mere, a.2)
Similarly, comparative and superlative uses are also predominantly attested for the meaning ‘insignificant, ordinary; inadequate, feeble’ (cf. (30)), but it is intuitively plausible that both mere ‘unmixed’ and mere ‘morally good’ could be used in comparison, too. (30)
I have never seen people that seemed merer animals. (OED: 1841, s.v. mere, a.2)
Furthermore, mere was used as an intensifier in two different ways: Firstly, it is attested as a neutral/positive intensifier, often when used with evaluatively neutral or positive nouns; here its meaning can be paraphrased with ‘nothing else/less than’. This use of mere is attested from late ME to EModE. (31) (32)
(33)
For no desyre of mans prayse or ytch of vain glory, but of mere humilitie. (OED: 1532, s.v. itch, n.) I am glad that yowe have the use of my brother Killegrewe, who, as he is of great experience, so I knowe he doth of verie, meere affection towardes your lordship serve theare at this time, which otherwise noe reward could provoke him, such desire I knowe he hath to live privatelie and unoccupied. (PCEEC: 1586, leycest) I am enforsed to trowble yow with theise ragged lynes and by meere necessytie urged and compelled to crave your ayde towchinge my case. (PCEEC: 1576, bacon)
Secondly, it functions as negative intensifier or downtoner and can be paraphrased with ‘nothing better than’. The downtoner mere is first attested from late ME and corresponds to the contrasting uses of mere in PDE. (34)
I knowe that what euer..they esteme, as principall rules and pillers of their knowledge to be nothing else but meare fables and toyes. (OED: 1561, s.v. pillar, n.)
284
Elke Gehweiler
(35)
The effect of it was this: that ‘‘the King hath sent him to putt her Ma ¼ tie ¼ . in minde of the auncient confederacies betweene the Kings of Poland and England; that never a Monarche in Europe did willinglie neglect their frendship; that he had ever frendlie receaved her merchants and subjects of all qualitie; that she had suffered his to be spoiled without restitution, not for lacke of knowledge of the violencies, but out of meer injustice, not caring to minister remedie, notwithstanding many particular petitions and letters receaved (PCEEC: 1597, origin3)
3.2. Mere in diachronic corpora In order to find out how the PDE determiner has emerged from the adjective, data from two diachronic corpora was used. As the first intensifying use of mere is attested in the OED from the mid-15th century (cf. OED s.v. mere, a.2) I analyzed all instances of mere from between 1400 and 1600 in the OED quotations database and all instances of mere in the PCEEC, which contains letters from between 1410 and 1681. 3.2.1. Mere in the OED (1400–1600)6. no.
%
adjective ‘unmixed’
9
6.2
adjective ‘morally good’
11
7.5
adjective ‘(racially) unmixed’
11
7.5
mere motion (law)
9
6.2
mere right (law)
5
3.4
degree modifier ‘entirely’
3
2
intensifier ‘nothing else/less than’
17
11.6
intensifier ‘nothing better than’
78
53.4
unclassified
3
2
total
146
100
As an adjective with the meaning ‘unmixed’ mere typically modified SUBSTANCE-nouns such as wine, milk or ivory. Typical collocates of mere ‘morally good’ were nouns denoting 6
The quotations database of the OED was searched for the spelling variants mere, meere, meare, meire, mer, meer, mear, myere, meir, myre.
The grammaticalization of privative adjectives: the case of mere
285
human beings (e.g. maiden) and abstract nouns denoting human feelings, behaviour, character (e.g. truth, goodness, pity). It was further used with nouns denoting people and their language (the mere English, the mere Irish, mere Irishry) and as part of the fixed expressions mere motion and mere right from the language of law. In its use as degree modifier of adjectives it can be paraphrased as ‘entirely’. The majority of uses of mere in the OED, however, are intensifying uses: there are 17 instances of neutral/positive intensification and a further 78 instances of negative intensification. 3.2.2. Mere in the PCEEC (1410–1681)7. no.
%
adjective ‘(racially) unmixed’
1
1.9
mere motion (law)
1
1.9
intensifier ‘nothing else/less than’
30
55.6
intensifier ‘nothing better than’
19
35.2
unclassified
3
5.6
total
54
100
In the PCEEC, too, the intensifying uses account for the majority of occurrences of mere. Here, mere is even almost exclusively used as intensifier, whereas the other meanings are hardly attested. This is most likely due to the make up of the corpus, which consists of personal letters (defined as letters written by an identifiable individual to another identifiable individual, including private letters (cf. Nurmi, 1999: 57)), which naturally contain many subjective forms. Here the neutral/positive intensifying uses are more frequent than the downtoner uses, although the line between the different intensifying uses is sometimes hard to draw.8
7
The PCEEC was searched for the spelling variants mere, meere, meare, meire, mer, meer, mear, myere, meir, myre. 8 I agree with the anonymous referee who has noted that it would be interesting to check whether the intensifying uses have increased over the 300-year period studied. But as the OED quotations database greatly varies with regard to the amount of text it contains for different time periods, and as the PCEEC is not representative for the general language, I will for now restrict myself to the two ‘‘snapshots’’ presented above.
286
Elke Gehweiler
3.3. The development of mere as an intensifier The emergence of the PDE intensifier mere from the adjective meaning ‘unmixed’ involved two steps. First, the adjective was extended to be used to describe human beings, their feelings, their behaviour etc., i.e. the ‘inner life’ of a person was conceptualized as a substance: a ‘pure’ person was morally good, honest, noble, free from sin. This metaphor came to be conventionalized gradually, and finally mere had acquired the new meaning ‘morally good’. In a second step, the intensifiers ‘nothing else/less than’ and ‘nothing better than’ evolved from the adjective mere ‘morally good’.9 A number of scholars have noted that some adjectives are ambiguous when used in attributive position and acquire a second reading they do not have when used predicatively (cf. e.g. Bolinger, 1967; cf. Quirk et al., 1985; Huddleston and Pullum, 2005). Bolinger (1967), for example calls this second reading ‘reference modification reading’, Quirk et al. (1985) ‘noninherent reading’. The most frequently cited example is old in old friend, which can mean ‘friend who is old, i.e. has lived for a long time’ (referent modification or inherent reading corresponding to reading in predicative use) or ‘long time friend’ (reference modification or noninherent reading, not possible when used predicatively).10 Other examples include big eater (‘eater who is big’ vs. ‘someone who eats a lot’), perfect stranger (‘a stranger who is perfect’ vs. ‘an absolute stranger’), or criminal lawyer (‘lawyer who is criminal’ vs. ‘lawyer who practises
9
An anonymous referee has noted that it intuitively appeared more likely that the intensifier reading had evolved directly from the ‘unmixed’ sense. As none of the corpus occurrences are ambiguous between these two readings, however, I would argue that the ‘‘detour’’ via the meaning ‘morally good’ was a necessary step in the evolution of the intensifiers. This is also confirmed by the largely parallel development of German lauter, discussed in Gehweiler (to appear), for which we also have a number of similar ambiguous examples. A further scenario, that does not involve ambiguous examples, however, could be the following: mere (in its original meaning ‘unmixed’) comes to be used with nouns denoting inner states etc. and directly acquires a reference modification reading, a positive reference modification reading when used with positive nouns or in positive contexts, a negative reference modification reading when used with negative nouns or in negative contexts. 10 A preliminary corpus analysis of old friend has yielded that it is never used to mean ‘friend who has lived for a long time’ (referent modification reading). (Note that old friend can have a second reference modification reading, ‘former friend’.) Similarly criminal lawyer never occurs in the corpus with the meaning ‘lawyer who is criminal’, big eater does not mean ‘eater who is big’, and perfect stranger is not used with the meaning ‘stranger who is perfect’, i.e. it seems that at least for some adjectives the referent modification reading in attributive position does not exist. However, this remains to be – and will be – studied for a larger number of adjectives that are said to be ambiguous between the two readings. This hypothesis could also lend some support to the second scenario proposed in footnote 9: if referent modification does not exist, then we would not have to assume ambiguous examples in the evolution of the intensifiers.
The grammaticalization of privative adjectives: the case of mere
287
criminal law’) (cf. Huddleston and Pullum, 2005: 553 for an overview of different types of noninherent reading). Privative adjectives such as mere can also have a noninherent or reference modification reading when used attributively. In some attributive uses mere acquires an intensifier reading (corresponding to the reference modification reading) in addition to its original meaning ‘pure’ (which would be the referent modification reading). This intensifier/referent modification reading is paraphrasable as ‘what is denoted by N in the full sense of the term’, expressing that the property expressed by the head noun applies to a maximum degree (cf. Huddleston and Pullum, 2005: 553). When modifying a neutral or positive noun, attributive mere can become ambiguous between meaning ‘morally good’ and a neutral or positive intensifying reading ‘nothing else/less than’, as in (36), where ‘‘mere truth’’ can be interpreted to mean ‘morally good, i.e. honest, sincere truth’ or ‘truth in the full sense of the term, i.e. nothing less than the truth’. (36)
We..wyll se yf..he wyll of hys awne mynde confesse the mere trawthe. (OED: 1536, s.v. mere, a.2)
When modifying a negative noun, the reference modification reading becomes negative; in the examples in (37) and (38) ‘‘mere creature’’ and ‘‘mere man’’ could be ambiguous between ‘morally good creature/man’ and ‘creature/man in the full sense of the term, i.e. nothing better than a creature/man (as opposed to a holy being)’, although an intensifier interpretation seems more likely. (37)
(38)
Some affirme the holy Ghost to be but a meere creature, as did Arius, . . . the Tropickes (OED: 1585–7, s.v. tropic, n. and a.1) Nestorius..held, that the Virgin..did not bring forth the sonne of God, but a sole and a mere man. (OED: 1586, s.v. mere, a.2)
Interestingly, a number of the early occurrences of mere ‘nothing better than’ contain a contrast, where something wordly is contrasted with something holy, as in example (38) above, where ‘‘sonne of God’’ is opposed to ‘‘mere man’’. In such uses, mere always modifies the inferior, i.e. the ‘‘wordly’’ part of the comparison. In Gehweiler (to appear) the relevance of such contrasting contexts for the evolution of the Early New High German intensifier lauter is discussed, which has also emerged from a privative adjective (cf. also Eckardt, 2006). Note that in these contrasting examples the negative meaning of mere can still be cancelled; mere can be interpreted to mean ‘morally good’. The contrasting contexts must therefore be regarded as bridging contexts in the sense of Heine (2002: 86): They ‘‘giv[e] rise to an inference in favour of a new meaning’’ so that the ‘‘target meaning [is] foregrounded’’.
288
Elke Gehweiler
Uses of mere like the ones in the above contexts, I would argue, triggered a reanalysis of mere: it came to be interpreted to mean ‘nothing better than’ and could accordingly be used with this unambiguously negative meaning. (39) (40)
It is a meere vanity and foolish braggry. (OED: 1576, s.v. braggery) I knowe that what euer..they esteme, as principall rules and pillers of their knowledge to be nothing else but meare fables and toyes. (OED: 1561, s.v. pillar, n.)
In these new uses, called switch contexts by Heine (2002: 86), the negative meaning could no longer be cancelled, mere could no longer be interpreted to mean ‘morally good’, because otherwise the semantics of mere and the semantics of the head noun or the semantics of the sentences would have overtly contradicted each other, as in (39) and (40). The switch contexts are ‘‘incompatible with the source meaning’’ (Heine, 2002: 86), i.e. in these contexts the meaning ‘nothing better than’ is conventionalized.
4. CONCLUSION The intensifier mere fulfils only one of the defining properties for adjectives, use in attributive position. Although superlative uses of mere are attested, these are partly limited to SSN constructions. A synchronic corpus analysis of mere has shown that mere is used in a restricted number of ways in PDE. These uses have in common that through the use of mere a contrast of some kind is established, where two items of the same kind are compared. This is in line with mere’s function as downtoner which requires gradability of some sort. The development from mere ‘unmixed’ to the PDE intensifier was said to involve grammaticalization and subjectification. Mere occurs as the first modifier in the premodifying string, which indicates a proximity to the word class determiner and is typical of adjectives which have undergone subjectification. Its new meaning is no longer referential and it is used to convey speaker evaluation. The emergence of the intensifier took place in two steps: First, the metaphorical meaning ‘morally good’ was conventionalized. The second step involved the reanalysis of mere ‘morally good’ when used attributively. This reanalysis was possible due to the ambiguity of mere in some uses: here a reference modification or noninherent reading ‘nothing better than’ was possible in addition to the meaning ‘morally good’. It is very likely that the reanalysis took place in contexts where a comparison interpretation (wordly vs. holy) could be inferred. When this new meaning was extended to unambiguous contexts, the new meaning of mere was conventionalized. The development of the privative adjective mere into an intensifier is not unique. The German privative adjectives, lauter and eitel, whose original meaning was also ‘pure, unmixed’ developed
The grammaticalization of privative adjectives: the case of mere
289
the same negative intensifying reading ‘nothing better than’ in late Middle High German/early Early New High German. The first ambiguous uses of lauter (for eitel the corpus data is not as comprehensive) are similarly attested in contrasting contexts with a wordly/holy comparison. In contrast to mere, however, lauter developed into a determiner, i.e. its grammaticalization is further advanced.
ACKNOWLEDGMENTS I would like to thank Christiane Fellbaum who made very helpful suggestions to improve this paper. Furthermore I thank the audience of the workshop ‘Diachronic Semantics and Pragmatics’ at ICHL 2007 for a number of valuable comments. Financial support of the SFB 447 funded by the Deutsche Forschungsgemeinschaft (DFG) is gratefully acknowledged.
REFERENCES Adamson, S. (2000). ‘‘A lovely little example. Word order options and category shift in the premodifying string’’, in O. Fischer, A. Rosenbach and D. Stein (eds.), Pathways of change: Grammaticalization in English. Amsterdam: Benjamins, 39–66. Biber, D., S. Johansson, G. Leech, S. Conrad and E. Finegan (1999). Longman grammar of spoken and written English. London: Longman. Bolinger, D. (1967). ‘‘Adjectives in English: Attribution and predication’’. Lingua 18: 1–34. Breban, T. (2006). English adjectives of comparison: Lexical and grammaticalized uses, PhD thesis, Katholieke Universiteit Leuven. Brems, L. (2007). ‘‘The grammaticalization of small size nouns. Reconsidering frequency and analogy’’. Journal of English Linguistics 35: 293–342. Brinton, L. and E. C. Traugott (2005). Lexicalization and language change. Cambridge: Cambridge University Press. Denison, D. (2006). ‘‘Category change and gradience in the determiner system’’, in A. van Kemenade and B. Los (eds.), The handbook of the history of English. Oxford: Blackwell, 279–304. Eckardt, R. (2006). Meaning change in grammaticalization. An inquiry into semantic reanalysis. Oxford: Oxford University Press. Gehweiler, E. (2009). ‘‘Mere mortals, bare essentials: A corpus-based analysis of two intensifiers’’, in C. Reinfandt and L. Eckstein (eds.), Anglistentag 2008 Tu¨bingen: Proceedings, WVT, Trier. Gehweiler, E. (to appear). ‘‘The grammaticalization of German lauter and eitel’’, in K. Stathi, E. Gehweiler and E. Ko¨nig (eds.), What’s new in grammaticalization? Amsterdam: Benjamins. Halliday, M. A. K. (1985). An introduction to functional grammar. London: Edward Arnold. Heine, B. (2002). ‘‘On the role of context in grammaticalization’’, in I. Wischer and G. Diewald (eds.), New reflections on grammaticalization. Amsterdam: Benjamins, 83–101. Hopper, P. and E. C. Traugott (2003). Grammaticalization. Cambridge: Cambridge University Press.
290
Elke Gehweiler
Huddleston, R. and G. Pullum (2005). The Cambridge grammar of the English language. Cambridge: Cambridge University Press. Leisi, E. (1967[1961]). Der Wortinhalt. Seine Struktur im Deutschen und Englischen, 3., durchgesehene und erweiterte Auflage, Quelle & Meyer, Heidelberg. Nurmi, A. (1999). ‘‘The corpus of early English correspondence sampler (CEECS)’’. ICAME Journal 23: 53–64. Quirk, R., S. Greenbaum, G. Leech and J. Svartvik (1985). A comprehensive grammar of the English language. London: Longman. Sinclair, J. (1991). Corpus, concordance, collocation. Oxford: Oxford University Press. Traugott, E. C. (2007). ‘‘The concepts of constructional mismatch and type-shifting from the perspective of grammaticalization’’. Cognitive Linguistics 18 (4): 523–557. Wierzbicka, A. (1986). ‘‘Precision in vagueness. The semantics of English ‘approximatives’’’. Journal of Pragmatics 10: 597–614.
CORPORA BNC: British National Corpus, accessed with Sketch Engine. www.sketchengine.co.uk OED: Oxford English Dictionary, online version. www.oed.com PCEEC: Parsed Corpus of Early English Correspondence, parsed version. (2006). Annotated by Ann Taylor, Arja Nurmi, Anthony Warner, Susan Pintzuk, and Terttu Nevalainen. Compiled by the CEEC Project Team. York: University of York and Helsinki: University of Helsinki. Distributed through the Oxford Text Archive.
Current Trends in Diachronic Semantics and Pragmatics Edited by Maj-Britt Mosegaard Hansen and Jacqueline Visconti r 2009 by Emerald Group Publishing Limited. All rights reserved.
15 THE ORIGIN OF SEMANTIC CHANGE IN DISCOURSE TRADITION: A CASE STUDY Katerina Stathi
1. INTRODUCTION This paper presents a case study of semantic change that starts out in a particular discourse tradition. It discusses the semantic development of the German verb geho¨ren and shows that in a construction with the passive perfect participle the verb has developed a deontic meaning in present-day German. This meaning of obligation or necessity preserves traces of the discourse tradition where the semantic development started, administrative and legal contexts. In order to describe this phenomenon, the notion of persistence (Hopper, 1991) will be elaborated. The literal meaning of geho¨ren is ‘belong, pertain’. The construction under consideration is illustrated in (1a) and (1b): (1)
a. Sportler, die des Dopings u¨berfu¨hrt worden sind, geho¨ren gesperrt. (Frankfurter Rundschau 04.03.1998) ‘Athletes who have been accused of doping, should be/deserve to be barred.’ b. ‘‘Das LadenschluXgesetz geho¨rt abgeschafft’’ hieb es am Montag in einer Pressemitteilung des Wirtschaftsministeriums. (Frankfurter Allgemeine 26.10.1993) ‘‘‘The law determining the closing of shops should be abolished’’; this was stated in a press release of the ministry of economics on Monday.’
292
Katerina Stathi
In (1), geho¨ren occurs in a construction with a passive perfect participle (participle II) – gesperrt ‘barred’ and abgeschafft ‘abolished’ – and expresses necessity. This sense is also expressed in (2): (2)
a. Die Kinder geho¨ren um sieben Uhr ins Bett. ‘The children should be (belong) in bed by seven o’clock.’ b. Er geho¨rt hinter Schloss und Riegel. ‘He should be/deserves to be under lock and key.’
The verb geho¨ren thus expresses necessity in two constructions: 1. geho¨ren þ participle II (cf. (1)) 2. geho¨ren þ local expression, usually a Prepositional Phrase (cf. (2)) In constructions like (1), geho¨ren is considered to be grammaticalized to a modal passive auxiliary. The various treatments proposed with regard to its status as (semi-)auxiliary will not be considered here (cf. Reis, 1976; Askedal 1984, 2005; Engel, 1996; Zifonun et al., 1997; Eroms, 2000; Hentschel and Weydt, 2003). Instead, we will trace the diachronic semantic development of this verb in Section 2 and present a synchronic corpus-based study of the use of the construction geho¨ren þ participle II in contemporary German in Section 3. Section 4 provides a discussion of the findings in relation to the notion of persistence (Hopper, 1991) and Section 5 concludes.
2. A BRIEF HISTORY
OF GEHO¨REN
In this section the semantic development of geho¨ren is traced on the basis of the evidence presented in the Deutsches Wo¨rterbuch (DWB, 2004). The verb geho¨ren is attested from the eighth century when it had the meaning ‘hear forcefully’.1 Used with the dative the verb developed the sense ‘listen, obey’.2 By metonymy, the meaning ‘be a part/member of a clan, family etc.’ arose: When people were asked whom they obeyed, the inference arose that they were part of that family. The verb is then expanded to refer to (i) membership in communities of all sorts (not only kin) and the relations of a member to the whole in general, and by extension, (ii) possession of all sorts. According to the DWB, the verb developed a special sense in legal and administrative contexts, as shown in (3); it then passed with this nuance to general use.
1 2
This meaning is preserved in Geho¨r ‘hearing, ear’. This meaning is preserved in Gehorsam ‘obedience’.
The origin of semantic change in discourse tradition: a case study
(3)
293
die holz, wiesen, wasser und weiden mit gebrauchen [ . . . ], gehoren aufs gericht, sollen uf den rechten glockenschlag volgen. (14th century) ‘Those who make use of wood, meadows, water, and willows belong to the assembly, they shall follow the chimes.’
Although the use of the verb in (3) relates to categorization or membership – i.e. it states that citizens making use of wood, meadows, water, and willows constitute or are members of the assembly – a pragmatic inference of necessity of their membership in the assembly arises in this context. In (3) the modal sollen ‘shall’ þ infinitive (volgen ‘follow’) could be an elaboration of the previous deontic inference of geho¨ren. Those who are members of the assembly should gather upon hearing the bell. The development illustrated here must have emerged in the fourteenth century. Subsequently the verb is used to refer to the proper place where somebody belongs, which is suitable for him/her, where somebody or something should (ideally) be, but is not at the moment; cf. (4). Note that in (4) the deontic inference of geho¨ren is spelled out in the following explication by the use of the verb muss ‘must’. (4)
. . . ein mistha¨tiger geho¨rt da und da hin vor gericht, d. h. er muss von rechts wegen dort gerichtet werden . . . (DWB, s.v. geho¨ren) ‘ . . . a delinquent belongs somewhere in court, i.e. he must be judged there . . . ’
Summarizing, the verb geho¨ren most probably developed the meaning of obligation and necessity in the fourteenth century. This meaning has its origin in legal and administrative contexts and arose by pragmatic inferencing or strengthening (cf. Traugott, 1988; Traugott and Ko¨nig, 1991). The inference arises in the following way: Due to the inescapable force of administrative organization and law, stating that somebody belongs to a group of people or a place leads to the inference that s/he has no choice, that s/he must belong there. The co-occurrence of geho¨ren with the participle II (as well as the potential reanalysis of the former as an auxiliary) is much more recent, probably dating from the eighteenth or nineteenth century. For details of the structural development of the construction cf. Stathi (to appear).
3. SYNCHRONIC CORPUS-BASED ANALYSIS In order to investigate the use of the construction geho¨ren þ participle II, a corpus analysis was undertaken on the basis of a large corpus of German.3 Corpus studies are frequently associated 3
For the corpus-based synchronic study two subcorpora of the Digitales Wo¨rterbuch der Deutschen Sprache (DWDS) were used (cf. www.dwds.de). First, the ‘‘DWDS core corpus’’, a reference corpus of twentieth century German consisting of 100 million tokens and containing five genres (journalism, literary texts, scientific texts, and other non-fiction) as well as a small part of spoken language. Second,
294
Katerina Stathi
Table 15.1. The nine most frequent participles in the geho¨ren-construction (all word forms of GEHO¨REN; N ¼ 738) Participles
Occurrences
abgeschafft ‘abolished’
88
verboten ‘forbidden’
55
bestraft ‘punished’
23
erschossen ‘shot’
10
eingesperrt ‘put in prison’
8
gea¨ndert ‘changed’
7
abgeschoben ‘deported’
6
abgewa¨hlt ‘voted out’
6
beseitigt ‘removed, eliminated’
6
with an amount of surprise concerning the results and frequently contradict intuitions (cf. Stubbs, 2001; Hoey, 2005). The corpus reveals that in German geho¨ren does not occur randomly with any participle in the construction at hand. By contrast, the majority of the participles that co-occur with geho¨ren belong to a small number of semantic classes. In the remaining part of this section we will concentrate on the semantic properties of the participles; in the next section we will ask what the semantic properties of the participles reveal about the meaning of geho¨ren. Tables 15.1 and 15.2 list the most frequently used participles.4 As can be seen in Tables 15.1 and 15.2, the most frequent participles refer to a negative or unpleasant action on the patient, which is the grammatical subject. In fact, this is true for 73% of all the participles that are attested in the corpus. Typical examples for the use of geho¨ren þ participle II are given in (5): (5)
a. Demonstranten und andere kritische Bu¨rger geho¨ren natu¨rlich kontrolliert, u¨berwacht und mit polizeilichen und juristischen MaXnahmen konfrontiert; in
the opportunistic ‘‘DWDS supplementary corpus’’ of about 900 million tokens, which has been compiled from electronic versions of daily and weekly newspapers of the 1990s. 4 The numbers given here relate to the word order geho¨ren þ participle II in main clauses; the order participle II þ geho¨ren which occurs in subordinate clauses has not been included in the counts but does not show differences.
The origin of semantic change in discourse tradition: a case study
295
Table 15.2. Quantitative distribution of the participles occurring in the geho¨ren-construction for the word forms geho¨ren, geho¨re, geho¨rst Participles abgeschafft ‘abolished’
Occurrences W20
verboten ‘forbidden’ beseitigt ‘abolished, eliminated’
W5
bestraft ‘punished’ vergast ‘gased’ geschlachtet ‘slaughtered’
2–5
ausgerottet ‘exterminated’ ausgewiesen ‘expelled’ geschlossen ‘closed (down)’ umgebracht ‘murdered’ erschossen ‘shot’ eingesperrt ‘put in prison’ abgelo¨st ‘replaced’ abgebaut ‘reduced’
der Regel reichen dafu¨r die bestehenden Gesetze aus, der Staat soll seine vorhandenen Mittel effektiv einsetzen. (Marxistische Zeit- und Streitschrift 01.01.1988) ‘Demonstrators and other critical citizens should of course be controlled, observed and confronted by police and lawful measures; usually the existing law suffices, the state should adopt the existing means more effectively.’ b. ‘‘Solche Kollegen geho¨ren vor die Tu¨r gesetzt’’, kommentierte zufrieden ein Gewerkschaftssekreta¨r, als das Berliner Landesarbeitsgericht im Mai 1991 die Ku¨ndigung eines Metzgermeisters besta¨tigte. Dab dieser seine Kolleginnen wiederholt in Busen und Po gekniffen hatte, werteten die Richter als ‘‘Sto¨rung des Betriebsfriedens’’. (die tageszeitung 05.03.1992) ‘‘‘Such colleagues should be thrown out’’, a trade union secretary commented with satisfaction, when the Berlin labour court confirmed the dismissal of a butcher
296
Katerina Stathi
master craftsman in May 1991. The fact that he repeatedly pinched his female colleagues’ bosoms and bottom was judged by the judges as ‘‘disturbing the peace in the company’’.’ c. Es sei ‘‘zum Kotzen’’, wenn den Arbeitslosen angesichts von vier Millionen Menschen ohne Job die Arbeitslosenhilfe geku¨rzt werde und man versuche, auch noch den alten Menschen die Sozialhilfe zu ku¨rzen. Diese Bundesregierung geho¨re ‘‘weggejagt’’. (Su¨ddeutsche Zeitung 22.02.1996) ‘It is ‘‘disgusting’’ that in the face of four million unemployed the unemployment assistance is cut down and it is also attempted to cut the older people’s social welfare. This federal government should be driven away.’
The participles that denote a negative or unpleasant action on the patient usually refer to the following actions: PUNISHMENT
Examples: bestraft ‘punished’, verhaftet ‘arrested’, abgeurteilt ‘condemned’, gefesselt ‘captivated’, eingesperrt ‘put in prison’ PHYSICAL OR VERBAL VIOLENCE
Examples: ausgepeitscht ‘whipped’, gesteinigt ‘stoned’, verpru¨gelt ‘battered’, geschlagen ‘beaten’, disqualifiziert ‘disqualified’, kritisiert ‘criticized’ CAUSING OF DEATH
Examples: abgeschossen ‘shot down’, erschossen ‘shot’, vergast ‘gased’, umgebracht ‘murdered’, totgeschlagen ‘killed’, aufgeha¨ngt ‘hung’, geschlachtet ‘slaughtered’, ausgerottet ‘exterminated’, hingerichtet ‘executed’ DESTRUCTION
Examples: bombardiert ‘bombed’, in die Luft gesprengt ‘blown’, abgerissen ‘demolished’, aufgelo¨st ‘suspended’, vernichtet ‘destroyed’, beseitigt ‘removed, eliminated’
actions of
RESTRICTION
Examples: eingeschra¨nkt ‘constricted’, beschra¨nkt ‘restricted’, abgebaut ‘reduced’, u¨berwacht ‘observed’, kontrolliert ‘controlled’, bedroht ‘threatened’, diffamiert ‘defamed’, geba¨ndigt ‘repressed’
actions of
REMOVAL
Examples: abgeschafft ‘abolished’, verboten ‘forbidden’, unterbunden ‘prevented’, geschlossen ‘closed’, eliminiert ‘eliminated’, weggejagt ‘driven away’, ausgewiesen ‘expelled’, verbannt ‘exiled’, rausgeworfen, rausgeschmissen ‘thrown out’, abgeschoben ‘deported’, abgewa¨hlt ‘voted out’, entlassen ‘fired’
actions of
EXCHANGE:
The origin of semantic change in discourse tradition: a case study
297
Examples: aussortiert ‘sorted out’, ausgemustert ‘sorted’, ausgemerzt ‘expunged’, weggesperrt ‘locked away’, ausgewechselt ‘replaced’, ausgetauscht ‘exchanged’, abgelo¨st ‘replaced’ What are the semantic properties of the remaining participles? (6) gives typical examples for participles that do not belong to the class expressing a negative or unpleasant action on the patient: (6) a. Eine derartige sta¨dtebauliche Neuorientierung [ . . . ] geho¨re breit o¨ffentlich diskutiert und mu¨sse Konsens bei ‘‘allen gesellschaftlichen Gruppen’’ sein. (Frankfurter Rundschau 14.06.2000) ‘Such a new urbanistic orientation should be discussed in public and must be the consensus among ‘‘all social groups’’.’ b. Auch offen verkaufte Eier, gerade offen verkaufte geho¨ren datiert! (die tageszeitung 27.04.1991) ‘Eggs that are sold unpacked, especially loosely sold eggs should be dated!’ Even though the participles in (6) do not denote a negative or unpleasant action, they share semantic features with the negative-impact participles. Both types of participles refer to actions that are relevant to public affairs and public opinion (cf. also Eroms, 2000).
4. DISCUSSION The synchronic corpus analysis revealed that there are strong preferences concerning the meaning of participles which occur in the construction geho¨ren þ participle II. The majority of participles refer to actions that are relevant to public affairs and public opinion and most of them express a negative or unpleasant action on the patient. It was also shown that these participles denote punishment of some sort. We can represent the usage conventions of this construction by a general pattern (cf. (7a)) and a subpattern (cf. (7b)): (7) a. geho¨ren þ [PPP[PUBLIC AFFAIRS, PUBLIC OPINION]] b. geho¨ren þ [PPP[NEGATIVE IMPACT ON PATIENT (PUNISHMENT)]] Since the subpattern (7b) accounts for 73% of the data we can claim that geho¨ren has negative semantic prosody (Louw, 1993) in this construction. Louw (1993: 157) defines semantic prosody as follows: A consistent aura of meaning with which a form is imbued by its collocates is referred to [ . . . ] as a semantic prosody. Semantic prosodies have been largely inaccessible to human intuition about language and they cannot be retrieved reliably through introspection.5
5
For criticism of the notion semantic prosody cf. Whitsitt (2005).
298
Katerina Stathi
The notion of semantic prosody is based on J. R. Firth’s view that phonetic features are present in a sound’s neighbors. Similarly, certain features of a word’s meaning can be found in its syntagmatic environment. The influence of a word’s meaning is spread to its surrounds and limits the choices available to the speaker. This fact is largely ignored both in dictionaries and grammars (Hoey, 2005: 22). Large electronic corpora and collocational analysis allow the precise determination of the collocates of words which might carry parts of the meaning of the target word. For example, the verb happen co-occurs predominantly with nouns that denote negative or unpleasant things; it has negative semantic prosody.6 Similarly, the collocates of geho¨ren show a remarkable semantic uniformity, as was shown in the previous section. This fact can be described as a semantic association between geho¨ren and the participles with which it co-occurs. For example, Hoey (2005: 24) states: Semantic association exists when a word or word sequence is associated in the mind of a language user with a semantic set or class, some members of which are also collocates for that user.
The verb geho¨ren together with the participle (i.e. the whole construction) can thus be seen as an extended lexical item in the spirit of Sinclair (1988) or as a construction in the sense of Construction Grammar (cf. Goldberg, 1995). In this lexical item or construction, negative-impact participles are conventionally associated with the verb. We want to claim that the semantic association of geho¨ren with the participles that denote a negative or unpleasant action (especially punishment) on the patient represents a type of persistence. Hopper (1991: 22) defines persistence as follows: When a form undergoes grammaticization from a lexical to a grammatical function, so long as it is grammatically viable some traces of its original lexical meanings tend to adhere to it, and details of its lexical history may be reflected in constraints on its grammatical distribution.
The construction geho¨ren þ participle II can be analyzed as a case of persistence due to the following reasons. First, although we did not focus on the (degree of) grammaticalization of geho¨ren, it was mentioned in Section 1 that geho¨ren is considered a (semi-)auxiliary in this particular construction (cf. also Stathi, to appear). Second, it is true that details of its lexical history strongly influence its distribution. More particularly, this is not manifested as a constraint, but as a strong preference, as was shown in the previous section. Nevertheless, in order to capture the particularities of this process, we propose an elaboration and extension of the notion of persistence. In the case of geho¨ren, it is not only traces of the verb’s
6 It has been observed that cases of negative semantic prosody far outnumber those of positive semantic prosody (Louw, 1993).
The origin of semantic change in discourse tradition: a case study
299
original lexical meaning(s) that tend to adhere to it, but traces of the discourse tradition or genre in which the meaning of necessity or obligation emerged. As was shown in Section 2, this meaning arose in administrative and legal contexts by pragmatic inference. The semantic preferences of the participles, which in their majority refer to actions of public concern, relate to and are a continuation of the discourse tradition which enabled the pragmatic inference in the first place. Hence it is observed that persistence may also include features of the context which triggered the semantic development instead of including only features of an earlier attested lexical meaning. It must again be emphasized that this insight cannot rely on intuitions and would not be possible to gain without collocational analysis since it is not the verb geho¨ren itself which is characterized by persistence, but the participle in the construction. Semantic change in a restricted contextual environment is also known from other cases of grammaticalization. For example, Moore (2006) shows that Latin videlicet developed to a discourse marker in English, but this use is restricted to legal documents (slander depositions). On the contrary, although the semantic change of geho¨ren started out in legal and administrative contexts, its use was extended to other contexts as well. In the preceding discussion, it was repeatedly stressed that the occurrence of geho¨ren with negative-impact participles referring to public matters is only a preference, not a restriction. The verb can be shown to occur also with other participles, albeit more infrequently; cf. (8): (8) a. Kaffee, Reis und Nudeln ‘‘geho¨ren eingetuppert’’, doziert Tupper-Beraterin Christa Berger vor dem Kaffeekra¨nzchen in Frau Werners kleiner Perlacher Wohnung. (Su¨ddeutsche Zeitung 30.03.1996, p. 15) ‘Coffee, rice and noodles should be put into tupperware, tupperware adviser Christa Berger lectures at the hen’s party in Mrs. Werner’s little flat in Perlach.’ b. Philosophische Probleme geho¨ren demnach naturalisiert [ . . . ]. (Frankfurter Allgemeine 08.03.2000, p. N6) ‘Philosophical problems should thus be naturalized [ . . . ].’ The participles in (8) do not semantically fit the general pattern (7a) or the subpattern (7b). Examples like (8) illustrate a weakening of persistence of the discourse context. At the moment this use is marginal. Although we cannot predict changes in meaning, in the case of geho¨ren subsequent development might involve loss of the strong connection with the original discourse tradition and the associated persistence characteristics and semantic preferences that were described here.
5. CONCLUSION In this paper it was shown how discourse tradition may lead to semantic change of a word or construction and how the discourse tradition may persist in the usage convention of this word or
300
Katerina Stathi
construction. This was shown by means of a case study, the construction geho¨ren þ participle II in German, in which the verb geho¨ren (literally ‘belong, pertain’) came to express necessity and obligation. In most grammars and syntactic treatments the verb geho¨ren is analyzed in isolation as a modal semi-auxiliary. Diachronic evidence reveals that the meaning of necessity and obligation arose in legal and administrative contexts via pragmatic inference. The meaning shift can be represented as follows: (9)
MEMBERSHIP IN A GROUP (ASSEMBLY, COURT)WOBLIGATORY MEMBERSHIP IN A GROUP
(due to pragmatic inference: force of the law and the obligation to participate in public matters) A synchronic corpus analysis showed that, in its present-day use, the construction geho¨ren þ participle II preserves features not only of the verb’s original lexical meaning, but also of the discourse tradition in which the meaning shift in (9) took place. The interesting point is that these features are not revealed if we look at the verb per se, but are preserved in its syntagmatic environment – this both determines and limits the choices of the speaker. It was shown that features of the source are most clearly visible in the meaning of the participles. The majority refers to public affairs, which reflect the original discourse context of law and administration. Second, most of them express a negative or unpleasant action on the patient, such as punishment. There is an obvious association of punishment with the existence of law and with the regulatory nature of administration. The example of geho¨ren reveals the importance of various factors in semantic and pragmatic change. First, it shows that the relevance of the context in semantic and pragmatic change goes beyond the local context and may also include the discourse tradition. Hence the role of discourse traditions (Koch and Oesterreicher, 1994), text types (Schank, 1984) or genres (Moore, 2006) should be reconsidered in historical semantics and pragmatics. In future research it might be important to embed microprocesses in their discourse context and consider the contact between subsystems such as the interplay between lexical meaning, sentence meaning, and textual meaning. In addition, the case of geho¨ren also requires the consideration of historical cultural information, which is indispensable if we wish to understand processes of diachronic semantics and pragmatics.
ACKNOWLEDGMENTS This research was carried out within the scope of the collaborative research center (SFB 447) ‘‘Kulturen des Performativen’’ at the Freie Universita¨t Berlin. Financial support is gratefully acknowledged. I would also like to thank Ekkehard Ko¨nig for his ideas and comments, the
The origin of semantic change in discourse tradition: a case study
301
audience at the Montreal workshop for the fruitful discussion as well as one anonymous reviewer for insightful comments and suggestions.
REFERENCES Askedal, J. O. (1984). ‘‘Grammatikalisierung und Auxiliarisierung im sogenannten bekommen/ kriegen/erhalten-Passiv des Deutschen’’. Kopenhagener Beitra¨ge zur Germanistischen Linguistik 22: 5–47. DWB ¼ Deutsches Wo¨rterbuch (2004). Von Jacob und Wilhelm Grimm (1854–1960): Der Digitale Grimm. Frankfurt: Zweitausendeins. Engel, U. (1996). Deutsche Grammatik. Heidelberg: Groos. Eroms, H.-W. (2000). Syntax der deutschen Sprache. Berlin: Walter de Gruyter. Goldberg, A. E. (1995). Constructions. A construction grammar approach to argument structure. Chicago: The University of Chicago Press. Hentschel, E. and H. Weydt (2003). Handbuch der deutschen Grammatik. 3rd Edition. Berlin: Walter de Gruyter. Hoey, M. (2005). Lexical priming. A new theory of words and language. London: Routledge. Hopper, P. (1991). ‘‘On some principles of grammaticalization’’, in E. C. Traugott and B. Heine (eds.), Approaches to grammaticalization. Amsterdam: John Benjamins, 18–35. Koch, P. and W. Oesterreicher (1994). ‘‘Schriftlichkeit und Sprache’’, in H. Gu¨nther and O. Ludwig (eds.), Schrift und Schriftlichkeit: Ein internationales Handbuch internationaler Forschung. Berlin: Walter de Gruyter, 587–604. Louw, B. (1993). ‘‘Irony in the text or sincerity in the writer? The diagnostic potential of semantic prosodies’’, in M. Baker, G. Francis and E. Tognini-Bonelli (eds.), Text and technology: In honour of John Sinclair. Philadelphia: John Benjamins, 157–176. Moore, C. (2006). ‘‘The use of videlicet in Early Modern slander depositions. A case of genre-specific grammaticalization’’. Journal of Historical Pragmatics 7: 245–263. Reis, M. (1976). ‘‘Zum grammatischen Status der Hilfsverben’’. Beitra¨ge zur Geschichte der deutschen Sprache und Literatur 98: 64–82. Schank, G. (1984). ‘‘Ansa¨tze zu einer Theorie des Sprachwandels auf der Grundlage von Textsorten’’, in W. Besch, A. Belten, O. Reichmann and S. Sonderegger (eds.), Sprachgeschichte: Ein Handbuch zur Geschichte der deutschen Sprache und ihrer Erforschung. Berlin: Walter de Gruyter, 761–768. Sinclair, J. (1988). ‘‘The lexical item’’, in E. Weigand (ed.), Contrastive lexical semantics. Amsterdam: John Benjamins, 1–24. Stathi, K. (to appear). ‘‘Is German geho¨ren an auxiliary? The grammaticalization of the construction geho¨ren þ participle II’’, in K. Stathi, E. Gehweiler and E. Ko¨nig (eds.), What’s new in grammaticalization? Amsterdam: John Benjamins. Stubbs, M. (2001). Words and phrases: Corpus studies of lexical semantics. Oxford: Blackwell. Traugott, E. C. (1988). ‘‘Pragmatic strengthening and grammaticalization’’, Proceedings of the Fourteenth Annual Meeting, February 13–15. General session and parasession on grammaticalization. Berkeley: Berkeley Linguistics Society, 406–416.
302
Katerina Stathi
Traugott, E. C. and E. Ko¨nig (1991). ‘‘The semantics-pragmatics of grammaticalization revisited’’, in E. C. Traugott and B. Heine (eds.), Approaches to grammaticalization. Amsterdam: John Benjamins, 189–218. Whitsitt, S. (2005). ‘‘A critique of the concept of semantic prosody’’. International Journal of Corpus Linguistics 10: 283–305. Zifonun, G., L. Hoffmann and B. Strecker, et al. (1997). Grammatik der deutschen Sprache. Berlin: Walter de Gruyter.